NFSv4 S. Shepler Internet-Draft M. Eisler Intended status: Standards Track D. Noveck Expires: September 5, 2007 Editors March 4, 2007 NFSv4 Minor Version 1 draft-ietf-nfsv4-minorversion1-10.txt Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on September 5, 2007. Copyright Notice Copyright (C) The IETF Trust (2007). Abstract This Internet-Draft describes NFSv4 minor version one, including features retained from the base protocol and protocol extensions made subsequently. The current draft includes description of the major extensions, Sessions, Directory Delegations, and parallel NFS (pNFS). This Internet-Draft is an active work item of the NFSv4 working group. Active and resolved issues may be found in the issue tracker at: http://www.nfsv4-editor.org/cgi-bin/roundup/nfsv4. New issues Shepler, et al. Expires September 5, 2007 [Page 1] Internet-Draft NFSv4 Minor Version 1 March 2007 related to this document should be raised with the NFSv4 Working Group nfsv4@ietf.org and logged in the issue tracker. Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [1]. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 10 1.1. The NFSv4.1 Protocol . . . . . . . . . . . . . . . . . . 10 1.2. NFS Version 4 Goals . . . . . . . . . . . . . . . . . . 10 1.3. Minor Version 1 Goals . . . . . . . . . . . . . . . . . 11 1.4. Overview of NFS version 4.1 Features . . . . . . . . . . 11 1.4.1. RPC and Security . . . . . . . . . . . . . . . . . . 12 1.4.2. Protocol Structure . . . . . . . . . . . . . . . . . 12 1.4.3. File System Model . . . . . . . . . . . . . . . . . 13 1.4.4. Locking Facilities . . . . . . . . . . . . . . . . . 14 1.5. General Definitions . . . . . . . . . . . . . . . . . . 15 1.6. Differences from NFSv4.0 . . . . . . . . . . . . . . . . 17 2. Core Infrastructure . . . . . . . . . . . . . . . . . . . . . 17 2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . 18 2.2. RPC and XDR . . . . . . . . . . . . . . . . . . . . . . 18 2.2.1. RPC-based Security . . . . . . . . . . . . . . . . . 18 2.3. COMPOUND and CB_COMPOUND . . . . . . . . . . . . . . . . 21 2.4. Client Identifiers and Client Owners . . . . . . . . . . 22 2.4.1. Server Release of Client ID . . . . . . . . . . . . 26 2.4.2. Handling Client Owner Conflicts . . . . . . . . . . 26 2.5. Server Owners . . . . . . . . . . . . . . . . . . . . . 27 2.6. Security Service Negotiation . . . . . . . . . . . . . . 27 2.6.1. NFSv4 Security Tuples . . . . . . . . . . . . . . . 28 2.6.2. SECINFO and SECINFO_NO_NAME . . . . . . . . . . . . 28 2.6.3. Security Error . . . . . . . . . . . . . . . . . . . 28 2.7. Minor Versioning . . . . . . . . . . . . . . . . . . . . 32 2.8. Non-RPC-based Security Services . . . . . . . . . . . . 34 2.8.1. Authorization . . . . . . . . . . . . . . . . . . . 34 2.8.2. Auditing . . . . . . . . . . . . . . . . . . . . . . 34 2.8.3. Intrusion Detection . . . . . . . . . . . . . . . . 35 2.9. Transport Layers . . . . . . . . . . . . . . . . . . . . 35 2.9.1. Required and Recommended Properties of Transports . 35 2.9.2. Client and Server Transport Behavior . . . . . . . . 35 2.9.3. Ports . . . . . . . . . . . . . . . . . . . . . . . 37 2.10. Session . . . . . . . . . . . . . . . . . . . . . . . . 37 2.10.1. Motivation and Overview . . . . . . . . . . . . . . 37 2.10.2. NFSv4 Integration . . . . . . . . . . . . . . . . . 38 Shepler, et al. Expires September 5, 2007 [Page 2] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.3. Channels . . . . . . . . . . . . . . . . . . . . . . 39 2.10.4. Exactly Once Semantics . . . . . . . . . . . . . . . 42 2.10.5. RDMA Considerations . . . . . . . . . . . . . . . . 51 2.10.6. Sessions Security . . . . . . . . . . . . . . . . . 53 2.10.7. Session Mechanics - Steady State . . . . . . . . . . 57 2.10.8. Session Mechanics - Recovery . . . . . . . . . . . . 59 2.10.9. Parallel NFS and Sessions . . . . . . . . . . . . . 62 3. Protocol Data Types . . . . . . . . . . . . . . . . . . . . . 62 3.1. Basic Data Types . . . . . . . . . . . . . . . . . . . . 62 3.2. Structured Data Types . . . . . . . . . . . . . . . . . 64 4. Filehandles . . . . . . . . . . . . . . . . . . . . . . . . . 73 4.1. Obtaining the First Filehandle . . . . . . . . . . . . . 74 4.1.1. Root Filehandle . . . . . . . . . . . . . . . . . . 74 4.1.2. Public Filehandle . . . . . . . . . . . . . . . . . 74 4.2. Filehandle Types . . . . . . . . . . . . . . . . . . . . 75 4.2.1. General Properties of a Filehandle . . . . . . . . . 75 4.2.2. Persistent Filehandle . . . . . . . . . . . . . . . 76 4.2.3. Volatile Filehandle . . . . . . . . . . . . . . . . 76 4.3. One Method of Constructing a Volatile Filehandle . . . . 77 4.4. Client Recovery from Filehandle Expiration . . . . . . . 78 5. File Attributes . . . . . . . . . . . . . . . . . . . . . . . 79 5.1. Mandatory Attributes . . . . . . . . . . . . . . . . . . 80 5.2. Recommended Attributes . . . . . . . . . . . . . . . . . 80 5.3. Named Attributes . . . . . . . . . . . . . . . . . . . . 81 5.4. Classification of Attributes . . . . . . . . . . . . . . 81 5.5. Mandatory Attributes - Definitions . . . . . . . . . . . 83 5.6. Recommended Attributes - Definitions . . . . . . . . . . 84 5.7. Time Access . . . . . . . . . . . . . . . . . . . . . . 94 5.8. Interpreting owner and owner_group . . . . . . . . . . . 95 5.9. Character Case Attributes . . . . . . . . . . . . . . . 97 5.10. Quota Attributes . . . . . . . . . . . . . . . . . . . . 97 5.11. mounted_on_fileid . . . . . . . . . . . . . . . . . . . 98 5.12. Directory Notification Attributes . . . . . . . . . . . 99 5.12.1. dir_notif_delay . . . . . . . . . . . . . . . . . . 99 5.12.2. dirent_notif_delay . . . . . . . . . . . . . . . . . 99 5.13. PNFS Attributes . . . . . . . . . . . . . . . . . . . . 99 5.13.1. fs_layout_type . . . . . . . . . . . . . . . . . . . 99 5.13.2. layout_alignment . . . . . . . . . . . . . . . . . . 99 5.13.3. layout_blksize . . . . . . . . . . . . . . . . . . . 100 5.13.4. layout_hint . . . . . . . . . . . . . . . . . . . . 100 5.13.5. layout_type . . . . . . . . . . . . . . . . . . . . 100 5.13.6. mdsthreshold . . . . . . . . . . . . . . . . . . . . 100 5.14. Retention Attributes . . . . . . . . . . . . . . . . . . 101 6. Access Control Lists . . . . . . . . . . . . . . . . . . . . 103 6.1. Goals . . . . . . . . . . . . . . . . . . . . . . . . . 103 6.2. File Attributes Discussion . . . . . . . . . . . . . . . 104 6.2.1. ACL Attribute . . . . . . . . . . . . . . . . . . . 104 6.2.2. dacl and sacl Attributes . . . . . . . . . . . . . . 115 Shepler, et al. Expires September 5, 2007 [Page 3] Internet-Draft NFSv4 Minor Version 1 March 2007 6.2.3. mode Attribute . . . . . . . . . . . . . . . . . . . 116 6.2.4. mode_set_masked Attribute . . . . . . . . . . . . . 116 6.3. Common Methods . . . . . . . . . . . . . . . . . . . . . 117 6.3.1. Interpreting an ACL . . . . . . . . . . . . . . . . 117 6.3.2. Computing a Mode Attribute from an ACL . . . . . . . 118 6.4. Requirements . . . . . . . . . . . . . . . . . . . . . . 119 6.4.1. Setting the mode and/or ACL Attributes . . . . . . . 120 6.4.2. Retrieving the mode and/or ACL Attributes . . . . . 121 6.4.3. Creating New Objects . . . . . . . . . . . . . . . . 122 7. Single-server Name Space . . . . . . . . . . . . . . . . . . 125 7.1. Server Exports . . . . . . . . . . . . . . . . . . . . . 126 7.2. Browsing Exports . . . . . . . . . . . . . . . . . . . . 126 7.3. Server Pseudo File System . . . . . . . . . . . . . . . 126 7.4. Multiple Roots . . . . . . . . . . . . . . . . . . . . . 127 7.5. Filehandle Volatility . . . . . . . . . . . . . . . . . 127 7.6. Exported Root . . . . . . . . . . . . . . . . . . . . . 127 7.7. Mount Point Crossing . . . . . . . . . . . . . . . . . . 128 7.8. Security Policy and Name Space Presentation . . . . . . 128 8. File Locking and Share Reservations . . . . . . . . . . . . . 129 8.1. Locking . . . . . . . . . . . . . . . . . . . . . . . . 130 8.1.1. Client and Session ID . . . . . . . . . . . . . . . 130 8.1.2. State-owner Definition . . . . . . . . . . . . . . . 130 8.1.3. Stateid Definition . . . . . . . . . . . . . . . . . 131 8.1.4. Use of the Stateid and Locking . . . . . . . . . . . 134 8.2. Lock Ranges . . . . . . . . . . . . . . . . . . . . . . 137 8.3. Upgrading and Downgrading Locks . . . . . . . . . . . . 137 8.4. Blocking Locks . . . . . . . . . . . . . . . . . . . . . 138 8.5. Lease Renewal . . . . . . . . . . . . . . . . . . . . . 138 8.6. Crash Recovery . . . . . . . . . . . . . . . . . . . . . 139 8.6.1. Client Failure and Recovery . . . . . . . . . . . . 139 8.6.2. Server Failure and Recovery . . . . . . . . . . . . 140 8.6.3. Network Partitions and Recovery . . . . . . . . . . 143 8.7. Server Revocation of Locks . . . . . . . . . . . . . . . 147 8.8. Share Reservations . . . . . . . . . . . . . . . . . . . 148 8.9. OPEN/CLOSE Operations . . . . . . . . . . . . . . . . . 149 8.10. Open Upgrade and Downgrade . . . . . . . . . . . . . . . 149 8.11. Short and Long Leases . . . . . . . . . . . . . . . . . 150 8.12. Clocks, Propagation Delay, and Calculating Lease Expiration . . . . . . . . . . . . . . . . . . . . . . . 151 8.13. Vestigial Locking Infrastructure From V4.0 . . . . . . . 151 9. Client-Side Caching . . . . . . . . . . . . . . . . . . . . . 152 9.1. Performance Challenges for Client-Side Caching . . . . . 153 9.2. Delegation and Callbacks . . . . . . . . . . . . . . . . 153 9.2.1. Delegation Recovery . . . . . . . . . . . . . . . . 155 9.3. Data Caching . . . . . . . . . . . . . . . . . . . . . . 157 9.3.1. Data Caching and OPENs . . . . . . . . . . . . . . . 157 9.3.2. Data Caching and File Locking . . . . . . . . . . . 158 9.3.3. Data Caching and Mandatory File Locking . . . . . . 160 Shepler, et al. Expires September 5, 2007 [Page 4] Internet-Draft NFSv4 Minor Version 1 March 2007 9.3.4. Data Caching and File Identity . . . . . . . . . . . 160 9.4. Open Delegation . . . . . . . . . . . . . . . . . . . . 161 9.4.1. Open Delegation and Data Caching . . . . . . . . . . 164 9.4.2. Open Delegation and File Locks . . . . . . . . . . . 165 9.4.3. Handling of CB_GETATTR . . . . . . . . . . . . . . . 165 9.4.4. Recall of Open Delegation . . . . . . . . . . . . . 168 9.4.5. Clients that Fail to Honor Delegation Recalls . . . 170 9.4.6. Delegation Revocation . . . . . . . . . . . . . . . 171 9.5. Data Caching and Revocation . . . . . . . . . . . . . . 171 9.5.1. Revocation Recovery for Write Open Delegation . . . 172 9.6. Attribute Caching . . . . . . . . . . . . . . . . . . . 173 9.7. Data and Metadata Caching and Memory Mapped Files . . . 175 9.8. Name Caching . . . . . . . . . . . . . . . . . . . . . . 177 9.9. Directory Caching . . . . . . . . . . . . . . . . . . . 178 10. Multi-Server Name Space . . . . . . . . . . . . . . . . . . . 179 10.1. Location attributes . . . . . . . . . . . . . . . . . . 179 10.2. File System Presence or Absence . . . . . . . . . . . . 179 10.3. Getting Attributes for an Absent File System . . . . . . 181 10.3.1. GETATTR Within an Absent File System . . . . . . . . 181 10.3.2. READDIR and Absent File Systems . . . . . . . . . . 182 10.4. Uses of Location Information . . . . . . . . . . . . . . 183 10.4.1. File System Replication . . . . . . . . . . . . . . 183 10.4.2. File System Migration . . . . . . . . . . . . . . . 185 10.4.3. Referrals . . . . . . . . . . . . . . . . . . . . . 186 10.5. Additional Client-side Considerations . . . . . . . . . 187 10.6. Effecting File System Transitions . . . . . . . . . . . 188 10.6.1. File System Transitions and Simultaneous Access . . 189 10.6.2. Simultaneous Use and Transparent Transitions . . . . 190 10.6.3. Filehandles and File System Transitions . . . . . . 192 10.6.4. Fileid's and File System Transitions . . . . . . . . 192 10.6.5. Fsids and File System Transitions . . . . . . . . . 193 10.6.6. The Change Attribute and File System Transitions . . 193 10.6.7. Lock State and File System Transitions . . . . . . . 194 10.6.8. Write Verifiers and File System Transitions . . . . 197 10.7. Effecting File System Referrals . . . . . . . . . . . . 197 10.7.1. Referral Example (LOOKUP) . . . . . . . . . . . . . 198 10.7.2. Referral Example (READDIR) . . . . . . . . . . . . . 202 10.8. The Attribute fs_absent . . . . . . . . . . . . . . . . 204 10.9. The Attribute fs_locations . . . . . . . . . . . . . . . 204 10.10. The Attribute fs_locations_info . . . . . . . . . . . . 206 10.10.1. The fs_locations_server4 Structure . . . . . . . . . 209 10.10.2. The fs_locations_info4 Structure . . . . . . . . . . 214 10.10.3. The fs_locations_item4 Structure . . . . . . . . . . 215 10.11. The Attribute fs_status . . . . . . . . . . . . . . . . 216 11. Directory Delegations . . . . . . . . . . . . . . . . . . . . 220 11.1. Introduction to Directory Delegations . . . . . . . . . 220 11.2. Directory Delegation Design . . . . . . . . . . . . . . 221 11.3. Attributes in Support of Directory Notifications . . . . 222 Shepler, et al. Expires September 5, 2007 [Page 5] Internet-Draft NFSv4 Minor Version 1 March 2007 11.4. Delegation Recall . . . . . . . . . . . . . . . . . . . 222 11.5. Directory Delegation Recovery . . . . . . . . . . . . . 222 12. Parallel NFS (pNFS) . . . . . . . . . . . . . . . . . . . . . 222 12.1. Introduction . . . . . . . . . . . . . . . . . . . . . . 222 12.2. PNFS Definitions . . . . . . . . . . . . . . . . . . . . 224 12.2.1. Metadata . . . . . . . . . . . . . . . . . . . . . . 224 12.2.2. Metadata Server . . . . . . . . . . . . . . . . . . 224 12.2.3. Client . . . . . . . . . . . . . . . . . . . . . . . 225 12.2.4. Storage Device . . . . . . . . . . . . . . . . . . . 225 12.2.5. Data Server . . . . . . . . . . . . . . . . . . . . 225 12.2.6. Storage Protocol or Data Protocol . . . . . . . . . 225 12.2.7. Control Protocol . . . . . . . . . . . . . . . . . . 225 12.2.8. Layout . . . . . . . . . . . . . . . . . . . . . . . 226 12.2.9. Layout Types . . . . . . . . . . . . . . . . . . . . 226 12.2.10. Layout Iomode . . . . . . . . . . . . . . . . . . . 226 12.2.11. Layout Segment . . . . . . . . . . . . . . . . . . . 227 12.2.12. Device IDs . . . . . . . . . . . . . . . . . . . . . 228 12.3. PNFS Operations . . . . . . . . . . . . . . . . . . . . 228 12.4. PNFS Attributes . . . . . . . . . . . . . . . . . . . . 229 12.5. Layout Semantics . . . . . . . . . . . . . . . . . . . . 229 12.5.1. Guarantees Provided by Layouts . . . . . . . . . . . 229 12.5.2. Getting a Layout . . . . . . . . . . . . . . . . . . 230 12.5.3. Committing a Layout . . . . . . . . . . . . . . . . 231 12.5.4. Recalling a Layout . . . . . . . . . . . . . . . . . 234 12.5.5. Metadata Server Write Propagation . . . . . . . . . 240 12.6. PNFS Mechanics . . . . . . . . . . . . . . . . . . . . . 240 12.7. Recovery . . . . . . . . . . . . . . . . . . . . . . . . 241 12.7.1. Client Recovery . . . . . . . . . . . . . . . . . . 241 12.7.2. Dealing with Lease Expiration on the Client . . . . 242 12.7.3. Dealing with Loss of Layout State on the Metadata Server . . . . . . . . . . . . . . . . . . . . . . . 243 12.7.4. Recovery from Metadata Server Restart . . . . . . . 244 12.7.5. Operations During Metadata Server Grace Period . . . 246 12.7.6. Storage Device Recovery . . . . . . . . . . . . . . 246 12.8. Metadata and Storage Device Roles . . . . . . . . . . . 247 12.9. Security Considerations . . . . . . . . . . . . . . . . 248 13. PNFS: NFSv4.1 File Layout Type . . . . . . . . . . . . . . . 249 13.1. Session Considerations . . . . . . . . . . . . . . . . . 249 13.2. File Layout Definitions . . . . . . . . . . . . . . . . 251 13.3. File Layout Data Types . . . . . . . . . . . . . . . . . 251 13.4. Interpreting the File Layout . . . . . . . . . . . . . . 255 13.5. Sparse and Dense Stripe Unit Packing . . . . . . . . . . 257 13.6. Data Server Multipathing . . . . . . . . . . . . . . . . 259 13.7. Operations Issued to NFSv4.1 Data Servers . . . . . . . 259 13.8. COMMIT Through Metadata Server . . . . . . . . . . . . . 260 13.9. Global Stateid Requirements . . . . . . . . . . . . . . 261 13.10. The Layout Iomode . . . . . . . . . . . . . . . . . . . 261 13.11. Data Server State Propagation . . . . . . . . . . . . . 261 Shepler, et al. Expires September 5, 2007 [Page 6] Internet-Draft NFSv4 Minor Version 1 March 2007 13.11.1. Lock State Propagation . . . . . . . . . . . . . . . 262 13.11.2. Open-mode Validation . . . . . . . . . . . . . . . . 262 13.11.3. File Attributes . . . . . . . . . . . . . . . . . . 263 13.12. Data Server Component File Size . . . . . . . . . . . . 263 13.13. Recovery Considerations . . . . . . . . . . . . . . . . 264 13.14. Security Considerations for the File Layout Type . . . . 265 14. Internationalization . . . . . . . . . . . . . . . . . . . . 265 14.1. Stringprep profile for the utf8str_cs type . . . . . . . 266 14.2. Stringprep profile for the utf8str_cis type . . . . . . 268 14.3. Stringprep profile for the utf8str_mixed type . . . . . 269 14.4. UTF-8 Related Errors . . . . . . . . . . . . . . . . . . 271 15. Error Values . . . . . . . . . . . . . . . . . . . . . . . . 271 15.1. Error Definitions . . . . . . . . . . . . . . . . . . . 271 15.2. Operations and their valid errors . . . . . . . . . . . 285 15.3. Callback operations and their valid errors . . . . . . . 299 15.4. Errors and the operations that use them . . . . . . . . 300 16. NFS version 4.1 Procedures . . . . . . . . . . . . . . . . . 307 16.1. Procedure 0: NULL - No Operation . . . . . . . . . . . . 307 16.2. Procedure 1: COMPOUND - Compound Operations . . . . . . 308 17. NFS version 4.1 Operations . . . . . . . . . . . . . . . . . 313 17.1. Operation 3: ACCESS - Check Access Rights . . . . . . . 313 17.2. Operation 4: CLOSE - Close File . . . . . . . . . . . . 315 17.3. Operation 5: COMMIT - Commit Cached Data . . . . . . . . 317 17.4. Operation 6: CREATE - Create a Non-Regular File Object . 319 17.5. Operation 7: DELEGPURGE - Purge Delegations Awaiting Recovery . . . . . . . . . . . . . . . . . . . . . . . . 322 17.6. Operation 8: DELEGRETURN - Return Delegation . . . . . . 323 17.7. Operation 9: GETATTR - Get Attributes . . . . . . . . . 323 17.8. Operation 10: GETFH - Get Current Filehandle . . . . . . 325 17.9. Operation 11: LINK - Create Link to a File . . . . . . . 326 17.10. Operation 12: LOCK - Create Lock . . . . . . . . . . . . 327 17.11. Operation 13: LOCKT - Test For Lock . . . . . . . . . . 331 17.12. Operation 14: LOCKU - Unlock File . . . . . . . . . . . 332 17.13. Operation 15: LOOKUP - Lookup Filename . . . . . . . . . 334 17.14. Operation 16: LOOKUPP - Lookup Parent Directory . . . . 335 17.15. Operation 17: NVERIFY - Verify Difference in Attributes . . . . . . . . . . . . . . . . . . . . . . . 337 17.16. Operation 18: OPEN - Open a Regular File . . . . . . . . 338 17.17. Operation 19: OPENATTR - Open Named Attribute Directory . . . . . . . . . . . . . . . . . . . . . . . 352 17.18. Operation 21: OPEN_DOWNGRADE - Reduce Open File Access . 354 17.19. Operation 22: PUTFH - Set Current Filehandle . . . . . . 355 17.20. Operation 23: PUTPUBFH - Set Public Filehandle . . . . . 356 17.21. Operation 24: PUTROOTFH - Set Root Filehandle . . . . . 357 17.22. Operation 25: READ - Read from File . . . . . . . . . . 358 17.23. Operation 26: READDIR - Read Directory . . . . . . . . . 360 17.24. Operation 27: READLINK - Read Symbolic Link . . . . . . 364 17.25. Operation 28: REMOVE - Remove File System Object . . . . 365 Shepler, et al. Expires September 5, 2007 [Page 7] Internet-Draft NFSv4 Minor Version 1 March 2007 17.26. Operation 29: RENAME - Rename Directory Entry . . . . . 367 17.27. Operation 31: RESTOREFH - Restore Saved Filehandle . . . 369 17.28. Operation 32: SAVEFH - Save Current Filehandle . . . . . 370 17.29. Operation 33: SECINFO - Obtain Available Security . . . 370 17.30. Operation 34: SETATTR - Set Attributes . . . . . . . . . 374 17.31. Operation 37: VERIFY - Verify Same Attributes . . . . . 376 17.32. Operation 38: WRITE - Write to File . . . . . . . . . . 377 17.33. Operation 40: BACKCHANNEL_CTL - Backchannel control . . 382 17.34. Operation 41: BIND_CONN_TO_SESSION . . . . . . . . . . . 383 17.35. Operation 42: EXCHANGE_ID - Instantiate Client ID . . . 387 17.36. Operation 43: CREATE_SESSION - Create New Session and Confirm Client ID . . . . . . . . . . . . . . . . . . . 395 17.37. Operation 44: DESTROY_SESSION - Destroy existing session . . . . . . . . . . . . . . . . . . . . . . . . 405 17.38. Operation 45: FREE_STATEID - Free stateid with no locks . . . . . . . . . . . . . . . . . . . . . . . . . 406 17.39. Operation 46: GET_DIR_DELEGATION - Get a directory delegation . . . . . . . . . . . . . . . . . . . . . . . 407 17.40. Operation 47: GETDEVICEINFO - Get Device Information . . 412 17.41. Operation 48: GETDEVICELIST . . . . . . . . . . . . . . 413 17.42. Operation 49: LAYOUTCOMMIT - Commit writes made using a layout . . . . . . . . . . . . . . . . . . . . . . . . 414 17.43. Operation 50: LAYOUTGET - Get Layout Information . . . . 417 17.44. Operation 51: LAYOUTRETURN - Release Layout Information . . . . . . . . . . . . . . . . . . . . . . 420 17.45. Operation 52: SECINFO_NO_NAME - Get Security on Unnamed Object . . . . . . . . . . . . . . . . . . . . . 423 17.46. Operation 53: SEQUENCE - Supply per-procedure sequencing and control . . . . . . . . . . . . . . . . . 424 17.47. Operation 54: SET_SSV . . . . . . . . . . . . . . . . . 429 17.48. Operation 55: TEST_STATEID - Test stateids for validity . . . . . . . . . . . . . . . . . . . . . . . . 431 17.49. Operation 56: WANT_DELEGATION . . . . . . . . . . . . . 432 17.50. Operation 57: DESTROY_CLIENTID - Destroy existing client ID . . . . . . . . . . . . . . . . . . . . . . . 435 17.51. Operation 10044: ILLEGAL - Illegal operation . . . . . . 436 18. NFS version 4.1 Callback Procedures . . . . . . . . . . . . . 437 18.1. Procedure 0: CB_NULL - No Operation . . . . . . . . . . 437 18.2. Procedure 1: CB_COMPOUND - Compound Operations . . . . . 437 19. NFS version 4.1 Callback Operations . . . . . . . . . . . . . 439 19.1. Operation 3: CB_GETATTR - Get Attributes . . . . . . . . 439 19.2. Operation 4: CB_RECALL - Recall an Open Delegation . . . 441 19.3. Operation 5: CB_LAYOUTRECALL . . . . . . . . . . . . . . 442 19.4. Operation 6: CB_NOTIFY - Notify directory changes . . . 444 19.5. Operation 7: CB_PUSH_DELEG . . . . . . . . . . . . . . . 447 19.6. Operation 8: CB_RECALL_ANY - Keep any N delegations . . 448 19.7. Operation 9: CB_RECALLABLE_OBJ_AVAIL . . . . . . . . . . 451 19.8. Operation 10: CB_RECALL_SLOT - change flow control Shepler, et al. Expires September 5, 2007 [Page 8] Internet-Draft NFSv4 Minor Version 1 March 2007 limits . . . . . . . . . . . . . . . . . . . . . . . . . 452 19.9. Operation 11: CB_SEQUENCE - Supply callback channel sequencing and control . . . . . . . . . . . . . . . . . 453 19.10. Operation 12: CB_WANTS_CANCELLED . . . . . . . . . . . . 455 19.11. Operation 13: CB_NOTIFY_LOCK - Notify of possible lock availability . . . . . . . . . . . . . . . . . . . 456 19.12. Operation 10044: CB_ILLEGAL - Illegal Callback Operation . . . . . . . . . . . . . . . . . . . . . . . 457 20. Security Considerations . . . . . . . . . . . . . . . . . . . 458 21. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 458 21.1. Defining new layout types . . . . . . . . . . . . . . . 458 22. References . . . . . . . . . . . . . . . . . . . . . . . . . 459 22.1. Normative References . . . . . . . . . . . . . . . . . . 459 22.2. Informative References . . . . . . . . . . . . . . . . . 460 Appendix A. Acknowledgments . . . . . . . . . . . . . . . . . . 461 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 462 Intellectual Property and Copyright Statements . . . . . . . . . 464 Shepler, et al. Expires September 5, 2007 [Page 9] Internet-Draft NFSv4 Minor Version 1 March 2007 1. Introduction 1.1. The NFSv4.1 Protocol The NFSv4.1 protocol is a minor version of the NFSv4 protocol described in [2]. It generally follows the guidelines for minor versioning model laid in Section 10 of RFC 3530. However, it diverges from guidelines 11 ("a client and server that supports minor version X must support minor versions 0 through X-1"), and 12 ("no features may be introduced as mandatory in a minor version"). These divergences are due to the introduction of the sessions model for managing non-idempotent operations and the RECLAIM_COMPLETE operation. These two new features are infrastructural in nature and simplify implementation of existing and other new features. Making them optional would add undue complexity to protocol definition and implementation. NFSv4.1 accordingly updates the Minor Versioning guidelines (Section 2.7). NFSv4.1, as a minor version, is consistent with the overall goals for NFS Version 4, but extends the protocol so as to better meet those goals, based on experiences with NFSv4.0. In addition, NFSv4.1 has adopted some additional goals, which motivate some of the major extensions in minor version 1. 1.2. NFS Version 4 Goals The NFS version 4 protocol is a further revision of the NFS protocol defined already by versions 2 [17]] and 3 [18]. It retains the essential characteristics of previous versions: design for easy recovery, independent of transport protocols, operating systems and file systems, simplicity, and good performance. The NFS version 4 revision has the following goals: o Improved access and good performance on the Internet. The protocol is designed to transit firewalls easily, perform well where latency is high and bandwidth is low, and scale to very large numbers of clients per server. o Strong security with negotiation built into the protocol. The protocol builds on the work of the ONCRPC working group in supporting the RPCSEC_GSS protocol. Additionally, the NFS version 4 protocol provides a mechanism to allow clients and servers the ability to negotiate security and require clients and servers to support a minimal set of security schemes. Shepler, et al. Expires September 5, 2007 [Page 10] Internet-Draft NFSv4 Minor Version 1 March 2007 o Good cross-platform interoperability. The protocol features a file system model that provides a useful, common set of features that does not unduly favor one file system or operating system over another. o Designed for protocol extensions. The protocol is designed to accept standard extensions within a framework that enable and encourages backward compatibility. 1.3. Minor Version 1 Goals Minor version one has the following goals, within the framework established by the overall version 4 goals. o To correct significant structural weaknesses and oversights discovered in the base protocol. o To add clarity and specificity to areas left unaddressed or not addressed in sufficient detail in the base protocol. o To add specific features based on experience with the existing protocol and recent industry developments. o To provide protocol support to take advantage of clustered server deployments including the ability to provide scalable parallel access to files distributed among multiple servers. 1.4. Overview of NFS version 4.1 Features To provide a reasonable context for the reader, the major features of NFS version 4.1 protocol will be reviewed in brief. This will be done to provide an appropriate context for both the reader who is familiar with the previous versions of the NFS protocol and the reader that is new to the NFS protocols. For the reader new to the NFS protocols, there is still a set of fundamental knowledge that is expected. The reader should be familiar with the XDR and RPC protocols as described in [3] and [4]. A basic knowledge of file systems and distributed file systems is expected as well. This description of version 4.1 features will not distinguish those added in minor version one from those present in the base protocol but will treat minor version 1 as a unified whole. See Section 1.6 for a description of the differences between the two minor versions. Shepler, et al. Expires September 5, 2007 [Page 11] Internet-Draft NFSv4 Minor Version 1 March 2007 1.4.1. RPC and Security As with previous versions of NFS, the External Data Representation (XDR) and Remote Procedure Call (RPC) mechanisms used for the NFS version 4.1 protocol are those defined in [3] and [4]. To meet end- to-end security requirements, the RPCSEC_GSS framework [5] will be used to extend the basic RPC security. With the use of RPCSEC_GSS, various mechanisms can be provided to offer authentication, integrity, and privacy to the NFS version 4 protocol. Kerberos V5 will be used as described in [6] to provide one security framework. The LIPKEY and SPKM-3 GSS-API mechanisms described in [7] will be used to provide for the use of user password and client/server public key certificates by the NFS version 4 protocol. With the use of RPCSEC_GSS, other mechanisms may also be specified and used for NFS version 4.1 security. To enable in-band security negotiation, the NFS version 4.1 protocol has operations which provide the client a method of querying the server about its policies regarding which security mechanisms must be used for access to the server's file system resources. With this, the client can securely match the security mechanism that meets the policies specified at both the client and server. 1.4.2. Protocol Structure 1.4.2.1. Core Protocol Unlike NFS Versions 2 and 3, which used a series of ancillary protocols (e.g. NLM, NSM, MOUNT), within all minor versions of NFS version 4 only a single RPC protocol is used to make requests of the server. Facilities that had been separate protocols, such as locking, are now integrated within a single unified protocol. 1.4.2.2. Parallel Access Minor version one supports high-performance data access to a clustered server implementation by enabling a separation of metadata access and data access, with the latter done to multiple servers in parallel. Such parallel data access is controlled by recallable objects known as "layouts", which are integrated into the protocol locking model. Clients direct requests for data access to a set of data servers specified by the layout via a data storage protocol which may be NFSv4.1 or may be another protocol. Shepler, et al. Expires September 5, 2007 [Page 12] Internet-Draft NFSv4 Minor Version 1 March 2007 1.4.3. File System Model The general file system model used for the NFS version 4.1 protocol is the same as previous versions. The server file system is hierarchical with the regular files contained within being treated as opaque octet streams. In a slight departure, file and directory names are encoded with UTF-8 to deal with the basics of internationalization. The NFS version 4.1 protocol does not require a separate protocol to provide for the initial mapping between path name and filehandle. All file systems exported by a server are presented as a tree so that all file systems are reachable from a special per-server global root filehandle. This allows LOOKUP operations to be used to perform functions previously provided by the MOUNT protocol. The server provides any necessary pseudo file systems to bridge any gaps that arise due to unexported gaps between exported file systems. 1.4.3.1. Filehandles As in previous versions of the NFS protocol, opaque filehandles are used to identify individual files and directories. Lookup-type and create operations are used to go from file and directory names to the filehandle which is then used to identify the object to subsequent operations. The NFS version 4.1 protocol provides support for persistent filehandles, guaranteed to be valid for the lifetime of the file system object designated. In addition it provides support to servers to provide filehandles with more limited validity guarantees, called volatile filehandles. 1.4.3.2. File Attributes The NFS version 4.1 protocol has a rich and extensible attribute structure. Only a small set of the defined attributes are mandatory and must be provided by all server implementations. The other attributes are known as "recommended" attributes. One significant recommended file attribute is the Access Control List (ACL) attribute. This attribute provides for directory and file access control beyond the model used in NFS Versions 2 and 3. The ACL definition allows for specification of specific sets of permissions for individual users and groups. In addition, ACL inheritance allows propagation of access permissions and restriction down a directory tree as file system objects are created. One other type of attribute is the named attribute. A named Shepler, et al. Expires September 5, 2007 [Page 13] Internet-Draft NFSv4 Minor Version 1 March 2007 attribute is an opaque octet stream that is associated with a directory or file and referred to by a string name. Named attributes are meant to be used by client applications as a method to associate application-specific data with a regular file or directory. 1.4.3.3. Multi-server Namespace NFS Version 4.1 contains a number of features to allow implementation of namespaces that cross server boundaries and that allow and facilitate a non-disruptive transfer of support for individual file systems between servers. They are all based upon attributes that allow one file system to specify alternate or new locations for that file system. These attributes may be used together with the concept of absent file system which provide specifications for additional locations but no actual file system content. This allows a number of important facilities: o Location attributes may be used with absent file systems to implement referrals whereby one server may direct the client to a file system provided by another server. This allows extensive multi-server namespaces to be constructed. o Location attributes may be provided for present file systems to provide the locations of alternate file system instances or replicas to be used in the event that the current file system instance becomes unavailable. o Location attributes may be provided when a previously present file system becomes absent. This allows non-disruptive migration of file systems to alternate servers. 1.4.4. Locking Facilities As mentioned previously, NFS v4.1, is a single protocol which includes locking facilities. These locking facilities include support for many types of locks including a number of sorts of recallable locks. Recallable locks such as delegations allow the client to be assured that certain events will not occur so long as that lock is held. When circumstances change, the lock is recalled via a callback request. The assurances provided by delegations allow more extensive caching to be done safely when circumstances allow it. o Share reservations as established by OPEN operations. o Byte-range locks. Shepler, et al. Expires September 5, 2007 [Page 14] Internet-Draft NFSv4 Minor Version 1 March 2007 o File delegations which are recallable locks that assure the holder that inconsistent opens and file changes cannot occur so long as the delegation is held. o Directory delegations which are recallable delegations that assure the holder that inconsistent directory modifications cannot occur so long as the delegation is held. o Layouts which are recallable objects that assure the holder that direct access to the file data may be performed directly by the client and that no change to the data's location inconsistent with that access may be made so long as the layout is held. All locks for a given client are tied together under a single client- wide lease. All requests made on sessions associated with the client renew that lease. When leases are not promptly renewed lock are subject to revocation. In the event of server reinitialization, clients have the opportunity to safely reclaim their locks within a special grace period. 1.5. General Definitions The following definitions are provided for the purpose of providing an appropriate context for the reader. Client The "client" is the entity that accesses the NFS server's resources. The client may be an application which contains the logic to access the NFS server directly. The client may also be the traditional operating system client remote file system services for a set of applications. A client is uniquely identified by a Client Owner. In the case of file locking the client is the entity that maintains a set of locks on behalf of one or more applications. This client is responsible for crash or failure recovery for those locks it manages. Note that multiple clients may share the same transport and connection and multiple clients may exist on the same network node. Client ID A 64-bit quantity used as a unique, short-hand reference to a client supplied Verifier and client owner. The server is responsible for supplying the client ID. Shepler, et al. Expires September 5, 2007 [Page 15] Internet-Draft NFSv4 Minor Version 1 March 2007 Client Owner The client owner is a unique string, opaque to the server, which identifies a client. Multiple network connections and source network addresses originating those connections may share a client owner. The server is expected to treat requests from connnections with the same client owner has coming from the same client. Lease An interval of time defined by the server for which the client is irrevocably granted a lock. At the end of a lease period the lock may be revoked if the lease has not been extended. The lock must be revoked if a conflicting lock has been granted after the lease interval. All leases granted by a server have the same fixed interval. Note that the fixed interval was chosen to alleviate the expense a server would have in maintaining state about variable length leases across server failures. Lock The term "lock" is used to refer to any of record (octet-range) locks, share reservations, delegations or layouts unless specifically stated otherwise. Server The "Server" is the entity responsible for coordinating client access to a set of file systems. A server can span multiple network addresses. In NFSv4.1, a server is a two tiered entity allows for servers consisting of multiple components the flexibility to tightly or loosely couple their components without requiring tight synchronization among the components. Every server has a "Server Owner" which reflects the two tiers of a server entity. Server Owner The "Server Owner" identifies the server to the client. The server owner consists of a major and minor identifier. When the client has two connections each to a peer with the same major and minor identifier, the client assumes both peers are the same server (the server namespace is the same via each connection), and further assumes session and lock state is sharable across both connections. When each peer has the same major identifier but different minor identifier, the client assumes both peers can serve the same namespace, but session and lock state is not sharable across both connections. Stable Storage NFS version 4 servers must be able to recover without data loss from multiple power failures (including cascading power failures, that is, several power failures in quick succession), operating system failures, and hardware failure of components other than the storage medium itself (for example, disk, nonvolatile RAM). Shepler, et al. Expires September 5, 2007 [Page 16] Internet-Draft NFSv4 Minor Version 1 March 2007 Some examples of stable storage that are allowable for an NFS server include: 1. Media commit of data, that is, the modified data has been successfully written to the disk media, for example, the disk platter. 2. An immediate reply disk drive with battery-backed on- drive intermediate storage or uninterruptible power system (UPS). 3. Server commit of data with battery-backed intermediate storage and recovery software. 4. Cache commit with uninterruptible power system (UPS) and recovery software. Stateid A 128-bit quantity returned by a server that uniquely defines the open and locking state provided by the server for a specific open or lock owner for a specific file and type of lock. Verifier A 64-bit quantity generated by the client that the server can use to determine if the client has restarted and lost all previous lock state. 1.6. Differences from NFSv4.0 The following summarizes the differences between minor version one and the base protocol: o Implementation of the sessions model. o Support for parallel access to data. o Addition of the RECLAIM_COMPLETE operation to better structure the lock reclamation process. o Support for delegations on directories and other file types in addition to regular files. o Operations to re-obtain a delegation. o Support for client and server implementation id's. 2. Core Infrastructure Shepler, et al. Expires September 5, 2007 [Page 17] Internet-Draft NFSv4 Minor Version 1 March 2007 2.1. Introduction NFS version 4.1 (NFSv4.1) relies on core infrastructure common to nearly every operation. This core infrastructure is described in the remainder of this section. 2.2. RPC and XDR The NFS version 4.1 (NFSv4.1) protocol is a Remote Procedure Call (RPC) application that uses RPC version 2 and the corresponding eXternal Data Representation (XDR) as defined in RFC1831 [4] and RFC4506 [3]. 2.2.1. RPC-based Security Previous NFS versions have been thought of as having a host-based authentication model, where the NFS server authenticates the NFS client, and trust the client to authenticate all users. Actually, NFS has always depended on RPC for authentication. The first form of RPC authentication which required a host-based authentication approach. NFSv4 also depends on RPC for basic security services, and mandates RPC support for a user-based authentication model. The user-based authentication model has user principals authenticated by a server, and in turn the server authenticated by user principals. RPC provides some basic security services which are used by NFSv4. 2.2.1.1. RPC Security Flavors As described in section 7.2 "Authentication" of [4], RPC security is encapsulated in the RPC header, via a security or authentication flavor, and information specific to the specification of the security flavor. Every RPC header conveys information used to identify and authenticate a client and server. As discussed in Section 2.2.1.1.1, some security flavors provide additional security services. NFSv4 clients and servers MUST implement RPCSEC_GSS. (This requirement to implement is not a requirement to use.) Other flavors, such as AUTH_NONE, and AUTH_SYS, MAY be implemented as well. 2.2.1.1.1. RPCSEC_GSS and Security Services RPCSEC_GSS ([5]) uses the functionality of GSS-API RFC2743 [8]. This allows for the use of various security mechanisms by the RPC layer without the additional implementation overhead of adding RPC security flavors. Shepler, et al. Expires September 5, 2007 [Page 18] Internet-Draft NFSv4 Minor Version 1 March 2007 2.2.1.1.1.1. Identification, Authentication, Integrity, Privacy Via the GSS-API, RPCSEC_GSS can be used to identify and authenticate users on clients to servers, and servers to users. It can also perform integrity checking on the entire RPC message, including the RPC header, and the arguments or results. Finally, privacy, usually via encryption, is a service available with RPCSEC_GSS. Privacy is performed on the arguments and results. Note that if privacy is selected, integrity, authentication, and identification are enabled. If privacy is not selected, but integrity is selected, authentication and identification are enabled. If integrity and privacy are not selected, but authentication is enabled, identification is enabled. RPCSEC_GSS does not provide identification as a separate service. Although GSS-API has an authentication service distinct from its privacy and integrity services, GSS-API's authentication service is not used for RPCSEC_GSS's authentication service. Instead, each RPC request and response header is integrity protected with the GSS-API integrity service, and this allows RPCSEC_GSS to offer per-RPC authentication and identity. See [5] for more information. NFSv4 client and servers MUST support RPCSEC_GSS's integrity and authentication service. NFSv4.1 servers MUST support RPCSEC_GSS's privacy service. 2.2.1.1.1.2. Security mechanisms for NFS version 4 RPCSEC_GSS, via GSS-API, normalizes access to mechanisms that provide security services. Therefore NFSv4 clients and servers MUST support three security mechanisms: Kerberos V5, SPKM-3, and LIPKEY. The use of RPCSEC_GSS requires selection of: mechanism, quality of protection (QOP), and service (authentication, integrity, privacy). For the mandated security mechanisms, NFSv4 specifies that a QOP of zero (0) is used, leaving it up to the mechanism or the mechanism's configuration to use an appropriate level of protection that QOP zero maps to. Each mandated mechanism specifies minimum set of cryptographic algorithms for implementing integrity and privacy. NFSv4 clients and servers MUST be implemented on operating environments that comply with the mandatory cryptographic algorithms of each mandated mechanism. 2.2.1.1.1.2.1. Kerberos V5 The Kerberos V5 GSS-API mechanism as described in RFC1964 [6] ( [[Comment.1: need new Kerberos RFC]] ) MUST be implemented with the RPCSEC_GSS services as specified in the following table: Shepler, et al. Expires September 5, 2007 [Page 19] Internet-Draft NFSv4 Minor Version 1 March 2007 column descriptions: 1 == number of pseudo flavor 2 == name of pseudo flavor 3 == mechanism's OID 4 == RPCSEC_GSS service 5 == NFSv4.1 clients MUST support 6 == NFSv4.1 servers MUST support 1 2 3 4 5 6 ------------------------------------------------------------------ 390003 krb5 1.2.840.113554.1.2.2 rpc_gss_svc_none yes yes 390004 krb5i 1.2.840.113554.1.2.2 rpc_gss_svc_integrity yes yes 390005 krb5p 1.2.840.113554.1.2.2 rpc_gss_svc_privacy no yes Note that the number and name of the pseudo flavor is presented here as a mapping aid to the implementor. Because the NFSv4 protocol includes a method to negotiate security and it understands the GSS- API mechanism, the pseudo flavor is not needed. The pseudo flavor is needed for the NFS version 3 since the security negotiation is done via the MOUNT protocol as described in [19]. 2.2.1.1.1.2.2. LIPKEY The LIPKEY V5 GSS-API mechanism as described in [7] MUST be implemented with the RPCSEC_GSS services as specified in the following table: 1 2 3 4 5 6 ------------------------------------------------------------------ 390006 lipkey 1.3.6.1.5.5.9 rpc_gss_svc_none yes yes 390007 lipkey-i 1.3.6.1.5.5.9 rpc_gss_svc_integrity yes yes 390008 lipkey-p 1.3.6.1.5.5.9 rpc_gss_svc_privacy no yes 2.2.1.1.1.2.3. SPKM-3 as a security triple The SPKM-3 GSS-API mechanism as described in [7] MUST be implemented with the RPCSEC_GSS services as specified in the following table: 1 2 3 4 5 6 ------------------------------------------------------------------ 390009 spkm3 1.3.6.1.5.5.1.3 rpc_gss_svc_none yes yes 390010 spkm3i 1.3.6.1.5.5.1.3 rpc_gss_svc_integrity yes yes 390011 spkm3p 1.3.6.1.5.5.1.3 rpc_gss_svc_privacy no yes Shepler, et al. Expires September 5, 2007 [Page 20] Internet-Draft NFSv4 Minor Version 1 March 2007 2.2.1.1.1.3. GSS Server Principal Regardless of what security mechanism under RPCSEC_GSS is being used, the NFS server, MUST identify itself in GSS-API via a GSS_C_NT_HOSTBASED_SERVICE name type. GSS_C_NT_HOSTBASED_SERVICE names are of the form: service@hostname For NFS, the "service" element is nfs Implementations of security mechanisms will convert nfs@hostname to various different forms. For Kerberos V5, LIPKEY, and SPKM-3, the following form is RECOMMENDED: nfs/hostname 2.3. COMPOUND and CB_COMPOUND A significant departure from the versions of the NFS protocol before version 4 is the introduction of the COMPOUND procedure. For the NFSv4 protocol, in all minor versions, there are exactly two RPC procedures, NULL and COMPOUND. The COMPOUND procedure is defined as a series of individual operations and these operations perform the sorts of functions performed by traditional NFS procedures. The operations combined within a COMPOUND request are evaluated in order by the server, without any atomicity guarantees. A limited set of facilities exist to pass results from one operation to another. Once an operation returns a failing result, the evaluation ends and the results of all evaluated operations are returned to the client. With the use of the COMPOUND procedure, the client is able to build simple or complex requests. These COMPOUND requests allow for a reduction in the number of RPCs needed for logical file system operations. For example, multi-component lookup requests can be constructed by combining multiple LOOKUP operations. Those can be further combined with operations such as GETATTR, READDIR, or OPEN plus READ to do more complicated sets of operation without incurring additional latency. NFSv4 also contains a considerable set of callback operations in which the server makes an RPC directed at the client. Callback RPC's have a similar structure to that of the normal server requests. For the NFS version 4 protocol callbacks in all minor versions, there are two RPC procedures, NULL and CB_COMPOUND. The CB_COMPOUND procedure Shepler, et al. Expires September 5, 2007 [Page 21] Internet-Draft NFSv4 Minor Version 1 March 2007 is defined in an analogous fashion to that of COMPOUND with its own set of callback operations. Addition of new server and callback operation within the COMPOUND and CB_COMPOUND request framework provide means of extending the protocol in subsequent minor versions. Except for a small number of operations needed for session creation, server requests and callback requests are performed within the context of a session. Sessions provide a client context for every request and support robust replay protection for non-idempotent requests. 2.4. Client Identifiers and Client Owners For each operation that obtains or depends on locking state, the specific client must be determinable by the server. In NFSv4, each distinct client instance is represented by a client ID, which is a 64-bit identifier that identifies a specific client at a given time and which is changed whenever the client or the server re- initializes. Client IDs are used to support lock identification and crash recovery. In NFSv4.1, during steady state operation, the client ID associated with each operation is derived from the session (see Section 2.10) on which the operation is issued. Each session is associated with a specific client ID at session creation and that client ID then becomes the client ID associated with all requests issued using it. Therefore, unlike NFSv4.0, the only NFSv4.1 operations possible before a client ID is established, are those directly connected with establishing the client ID. A sequence of an EXCHANGE_ID operation followed by a CREATE_SESSION operation using that client ID (eir_clientid as returned from EXCHANGE_ID) is required to establish the identification on the server. Establishment of identification by a new incarnation of the client also has the effect of immediately releasing any locking state that a previous incarnation of that same client might have had on the server. Such released state would include all lock, share reservation, and, where the server is not supporting the CLAIM_DELEGATE_PREV claim type, all delegation state associated with same client with the same identity. For discussion of delegation state recovery, see Section 9.2.1. Releasing such state requires that the server be able to determine that one client instance is the successor of another. Where this cannot be done, for any of a number of reasons, the locking state will remain for a time subject to lease expiration (see Section 8.5) Shepler, et al. Expires September 5, 2007 [Page 22] Internet-Draft NFSv4 Minor Version 1 March 2007 and the new client will need to wait for such state to be removed, if it makes conflicting lock requests. Client identification is encapsulated in the following Client Owner structure: struct client_owner4 { verifier4 co_verifier; opaque co_ownerid; }; The first field, co_verifier, is a client incarnation verifier that is used to detect client reboots. Only if the co_verifier is different from that the server had previously recorded for the client (as identified by the second field of the structure, co_ownerid) does the server start the process of canceling the client's leased state. The second field, co_ownerid is a variable length string that uniquely defines the client so that subsequent instances of the same client bear the same co_ownerid with a different verifier. There are several considerations for how the client generates the co_ownerid string: o The string should be unique so that multiple clients do not present the same string. The consequences of two clients presenting the same string range from one client getting an error to one client having its leased state abruptly and unexpectedly canceled. o The string should be selected so the subsequent incarnations (e.g. reboots) of the same client cause the client to present the same string. The implementor is cautioned from an approach that requires the string to be recorded in a local file because this precludes the use of the implementation in an environment where there is no local disk and all file access is from an NFS version 4 server. o The string should be the same for each server network address that the client accesses, rather than common to all server network addresses (note: the precise opposite was advised in RFC3530). This way, if a server has multiple interfaces, the client can trunk traffic over multiple network paths as described in Section 2.10.3.4.1. o The algorithm for generating the string should not assume that the client's network address will not change, unless the client Shepler, et al. Expires September 5, 2007 [Page 23] Internet-Draft NFSv4 Minor Version 1 March 2007 implementation knows it is using statically assigned network addresses. This includes changes between client incarnations and even changes while the client is still running in its current incarnation. This means that if the client includes just the client's network address in the co_ownerid string, there is a real risk, with dynamic address assignment, that after the client gives up the network address, another client, using a similar algorithm for generating the co_ownerid string, would generate a conflicting co_ownerid string. Given the above considerations, an example of a well generated co_ownerid string is one that includes: o If applicable, the client's statically assigned network address. o Additional information that tends to be unique, such as one or more of: * The client machine's serial number (for privacy reasons, it is best to perform some one way function on the serial number). * A MAC address (again, a one way function should be performed). * The timestamp of when the NFS version 4 software was first installed on the client (though this is subject to the previously mentioned caution about using information that is stored in a file, because the file might only be accessible over NFS version 4). * A true random number. However since this number ought to be the same between client incarnations, this shares the same problem as that of the using the timestamp of the software installation. o For a user level NFS version 4 client, it should contain additional information to distinguish the client from other user level clients running on the same host, such as a process identifier or other unique sequence. As a security measure, the server MUST NOT cancel a client's leased state if the principal established the state for a given co_ownerid string is not the same as the principal issuing the EXCHANGE_ID. A server may compare an client_owner4 in a EXCHANGE_ID with an nfs_client_id4 established using SETCLIENTID using NFSv4 minor version 0, so that an NFSv4.1 client is not forced to delay until lease expiration for locking state established by the earlier client using minor version 0. This requires the client_owner4 be Shepler, et al. Expires September 5, 2007 [Page 24] Internet-Draft NFSv4 Minor Version 1 March 2007 constructed the same way as the nfs_client_id4. If the latter's contents included the server's network address, and the NFSv4.1 client does not wish to use a client ID that prevents trunking, it should issue two EXCHANGE_ID operations. The first EXCHANGE_ID will have a client_owner4 equal to the nfs_client_id4. This will clear the state created by the NFSv4.0 client. The second EXCHANGE_ID will not have the server's network address. The state created for the second EXCHANGE_ID will not have to wait for lease expiration, because there will be no state to expire. Once an EXCHANGE_ID has been done, and the resulting client ID established as associated with a session, all requests made on that session implicitly identify that client ID, which in turn designates the client specified using the long-form client_owner4 structure. The shorthand client identifier (a client ID) is assigned by the server (the eir_clientid result from EXCHANGE_ID) and should be chosen so that it will not conflict with a client ID previously assigned by the server. This applies across server restarts or reboots. In the event of a server restart, a client may find out that its current client ID is no longer valid when receives a NFS4ERR_STALE_CLIENTID error. The precise circumstances depend of the characteristics of the sessions involved, specifically whether the session is persistent (see Section 2.10.4.5). When a session is not persistent, the client will need to create a new session. When the existing client ID is presented to a server as part of creating a session and that client ID is not recognized, as would happen after a server reboot, the server will reject the request with the error NFS4ERR_STALE_CLIENTID. When this happens, the client must obtain a new client ID by use of the EXCHANGE_ID operation and then use that client ID as the basis of the basis of a new session and then proceed to any other necessary recovery for the server reboot case (See Section 8.6.2). In the case of the session being persistent, the client will re- establish communication using the existing session after the reboot. This session will be associated with a client ID that has had state revoked (but the persistent session is never associated with a stale client ID, because if the session is persistent, the client ID MUST persist), and the client will receive an indication of that fact in the sr_status_flags field returned by the SEQUENCE operation (see Section 17.46.4). The client can then use the existing session to do whatever operations are necessary to determine the status of requests outstanding at the time of reboot, while avoiding issuing new requests, particularly any involving locking on that session. Such requests would fail with an NFS4ERR_STALE_STATEID error, if Shepler, et al. Expires September 5, 2007 [Page 25] Internet-Draft NFSv4 Minor Version 1 March 2007 attempted. See the detailed descriptions of EXCHANGE_ID (Section 17.35 and CREATE_SESSION (Section 17.36) for a complete specification of these operations. 2.4.1. Server Release of Client ID NFSv4.1 introduces a new operation called DESTROY_CLIENTID (Section 17.50) which the client SHOULD use to destroy a client ID it no longer needs. This permits graceful, bilateral release of a client ID. If the server determines that the client holds no associated state for its client ID (including sessions, opens, locks, delegations, layouts, and wants), the server may choose to unilaterally release the client ID. The server may make this choice for an inactive client so that resources are not consumed by those intermittently active clients. If the client contacts the server after this release, the server must ensure the client receives the appropriate error so that it will use the EXCHANGE_ID/CREATE_SESSION sequence to establish a new identity. It should be clear that the server must be very hesitant to release a client ID since the resulting work on the client to recover from such an event will be the same burden as if the server had failed and restarted. Typically a server would not release a client ID unless there had been no activity from that client for many minutes. As long as there are sessions, opens, locks, delegations, layouts, or wants, the server MUST not release the client ID. See Section 2.10.8.1.4 for discussion on releasing inactive sessions. 2.4.2. Handling Client Owner Conflicts If the co_ownerid string in a EXCHANGE_ID request is properly constructed, and if the client takes care to use the same principal for each successive use of EXCHANGE_ID, then, barring an active denial of service attack, conflicts are not possible. However, client bugs, server bugs, or perhaps a deliberate change of the principal owner of the co_ownerid string (such as the case of a client that changes security flavors, and under the new flavor, there is no mapping to the previous owner) will in rare cases result in a conflict. When the server gets a EXCHANGE_ID for a client owner that currently has no state, or if it has state, but the lease has expired, server MUST allow the EXCHANGE_ID, and confirm the new client ID if followed by the appropriate CREATE_SESSION. Shepler, et al. Expires September 5, 2007 [Page 26] Internet-Draft NFSv4 Minor Version 1 March 2007 When the server gets a EXCHANGE_ID for a client owner that currently has state, or an unexpired lease, and the principal that issues the EXCHANGE_ID is different than principal the previously established the client owner, the server MUST not destroy the any state that currently exists for client owner. Regardless, the server has two choices. First, it can return NFS4ERR_CLID_INUSE. Second, it can allow the EXCHANGE_ID, and simply treat the client owner as consisting of both the co_ownerid and the principal that issued the EXCHANGE_ID. 2.5. Server Owners The Server Owner is somewhat similar to a Client Owner (Section 2.4), but unlike the Client Owner, there is no shorthand serverid. The Server Owner is defined in the following structure: struct server_owner4 { uint64_t so_minor_id; opaque so_major_id; }; The Server Owner is returned in the results of EXCHANGE_ID. When the so_major_id fields are the same in two EXCHANGE_ID results, the connections each EXCHANGE_ID are sent over can be assumed to address the same Server (as defined in Section 1.5). If the so_minor_id fields are also the same, then not only do both connections connect to the same server, but the session and other state can be shared across both connections. The reader is cautioned that multiple servers may deliberately or accidentally claim to have the same so_major_id or so_major_id/so_minor_id; the reader should examine Section 2.10.3.4.1 and Section 17.35. The considerations for generating an so_major_id are similar to that for generating a co_ownerid string (see Section 2.4). The consequences of two servers generating conflict so_major_id values are less dire than they are for co_ownerid conflicts because the client can use RPCSEC_GSS to compare the authenticity of each server (see Section 2.10.3.4.1). 2.6. Security Service Negotiation With the NFS version 4 server potentially offering multiple security mechanisms, the client needs a method to determine or negotiate which mechanism is to be used for its communication with the server. The NFS server may have multiple points within its file system namespace that are available for use by NFS clients. These points can be considered security policy boundaries, and in some NFS Shepler, et al. Expires September 5, 2007 [Page 27] Internet-Draft NFSv4 Minor Version 1 March 2007 implementations are tied to NFS export points. In turn the NFS server may be configured such that each of these security policy boundaries may have different or multiple security mechanisms in use. The security negotiation between client and server must be done with a secure channel to eliminate the possibility of a third party intercepting the negotiation sequence and forcing the client and server to choose a lower level of security than required or desired. See Section 20 for further discussion. 2.6.1. NFSv4 Security Tuples An NFS server can assign one or more "security tuples" to each security policy boundary in its namespace. Each security tuple consists of a security flavor (see Section 2.2.1.1), and if the flavor is RPCSEC_GSS, a GSS-API mechanism OID, a GSS-API quality of protection, and an RPCSEC_GSS service. 2.6.2. SECINFO and SECINFO_NO_NAME The SECINFO and SECINFO_NO_NAME operations allow the client to determine, on a per filehandle basis, what security tuple is to be used for server access. In general, the client will not have to use either operation except during initial communication with the server or when the client crosses security policy boundaries at the server. It is possible that the server's policies change during the client's interaction therefore forcing the client to negotiate a new security tuple. Where the use of different security tuples would affect the type of access that would be allowed if a request was issued over the same connection used for the SECINFO or SECINFO_NO_NAME operation (e.g. read-only vs. read-write) access, security tuples that allow greater access should be presented first. Where the general level of access is the same and different security flavors limit the range of principals whose privileges are recognized (e.g. allowing or disallowing root access), flavors supporting the greatest range of principals should be listed first. 2.6.3. Security Error Based on the assumption that each NFS version 4 client and server must support a minimum set of security (i.e., LIPKEY, SPKM-3, and Kerberos-V5 all under RPCSEC_GSS), the NFS client will initiate file access to the server with one of the minimal security tuples. During communication with the server, the client may receive an NFS error of NFS4ERR_WRONGSEC. This error allows the server to notify the client that the security tuple currently being used is contravenes the Shepler, et al. Expires September 5, 2007 [Page 28] Internet-Draft NFSv4 Minor Version 1 March 2007 server's security policy. The client is then responsible for determining (see Section 2.6.3.1) what security tuples are available at the server and choosing one which is appropriate for the client. 2.6.3.1. Using NFS4ERR_WRONGSEC, SECINFO, and SECINFO_NO_NAME This section explains of the mechanics of NFSv4.1 security negotiation. The term "put filehandle operation" refers to PUTROOTFH, PUTPUBFH, PUTFH, and RESTOREFH. 2.6.3.1.1. Put Filehandle Operation + SAVEFH The client is saving a filehandle for a future RESTOREFH. The server MUST NOT return NFS4ERR_WRONG to either the put filehandle operation or SAVEFH. 2.6.3.1.2. Two or More Put Filehandle Operations For a series of N put filehandle operations, the server MUST NOT return NFS4ERR_WRONGSEC to the first N-1 put filehandle operations. The Nth put filehandle operation is handled as if it is the first in a series of operations, and the second in the series of operations is not a put filehandle operation. For example if the server received PUTFH, PUTROOTFH, LOOKUP, then the PUTFH is ignored for NFS4ERR_WRONGSEC purposes, and the PUTROOTFH, LOOKUP subseries is processed as according to Section 2.6.3.1.3. 2.6.3.1.3. Put Filehandle Operation + LOOKUP (or OPEN by Name) This situation also applies to a put filehandle operation followed by a LOOKUP or an OPEN operation that specifies a component name. In this situation, the client is potentially crossing a security policy boundary, and the set of security tuples the parent directory supports differ from those of the child. The server implementation may decide whether to impose any restrictions on security policy administration. There are at least three approaches (sec_policy_child is the tuple set of the child export, sec_policy_parent is that of the parent). a) sec_policy_child <= sec_policy_parent (<= for subset). This means that the set of security tuples specified on the security policy of a child directory is always a subset of that of its parent directory. Shepler, et al. Expires September 5, 2007 [Page 29] Internet-Draft NFSv4 Minor Version 1 March 2007 b) sec_policy_child ^ sec_policy_parent != {} (^ for intersection, {} for the empty set). This means that the security tuples specified on the security policy of a child directory always has a non empty intersection with that of the parent. c) sec_policy_child ^ sec_policy_parent == {}. This means that the set of tuples specified on the security policy of a child directory may not intersect with that of the parent. In other words, there are no restrictions on how the system administrator may set up these tuples. For a server to support approach (b) (when client chooses a flavor that is not a member of sec_policy_parent) and (c), the put filehandle operation must NOT return NFS4ERR_WRONGSEC in case of security mismatch. Instead, it should be returned from the LOOKUP (or OPEN by component name) that follows. Since the above guideline does not contradict approach (a), it should be followed in general. Even if approach (a) is implemented, it is possible for the security tuple used to be acceptable for the target of LOOKUP but not for the filehandles used in the put filehandle operation. The put filehandle operation could be a PUTROOTFH or PUTPUBFH, where the client cannot know the security tuples for the root or public filehandle. Or the security policy for the filehandle used by the put filehandle operation could have changed since the time the filehandle was obtained. Therefore, an NFSv4.1 server MUST NOT return NFS4ERR_WRONGSEC in response to the put filehandle operation if the operation is immediately followed by a LOOKUP or an OPEN by component name. 2.6.3.1.4. Put Filehandle Operation + LOOKUPP Since SECINFO only works its way down, there is no way LOOKUPP can return NFS4ERR_WRONGSEC without SECINFO_NO_NAME. SECINFO_NO_NAME solves this issue because via style SECINFO_STYLE4_PARENT, it works in the opposite direction as SECINFO. As with Section 2.6.3.1.3, the put filehandle operation must not return NFS4ERR_WRONGSEC whenever it is followed by LOOKUPP. If the server does not support SECINFO_NO_NAME, the client's only recourse is to issue the put filehandle operation, LOOKUPP, GETFH sequence of operations with every security tuple it supports. Regardless whether SECINFO_NO_NAME is supported, an NFSv4.1 server MUST NOT return NFS4ERR_WRONGSEC in response to a put filehandle operation if the operation is immediately followed by a LOOKUPP. Shepler, et al. Expires September 5, 2007 [Page 30] Internet-Draft NFSv4 Minor Version 1 March 2007 2.6.3.1.5. Put Filehandle Operation + SECINFO/SECINFO_NO_NAME A security sensitive client is allowed to choose a strong security tuple when querying a server to determine a file object's permitted security tuples. The security tuple chosen by the client does not have to be included in the tuple list of the security policy of the either parent directory indicated in the put filehandle operation, or the child file object indicated in SECINFO (or any parent directory indicated in SECINFO_NO_NAME). Of course the server has to be configured for whatever security tuple the client selects, otherwise the request will fail at RPC layer with an appropriate authentication error. In theory, there is no connection between the security flavor used by SECINFO or SECINFO_NO_NAME and those supported by the security policy. But in practice, the client may start looking for strong flavors from those supported by the security policy, followed by those in the mandatory set. The NFSv4.1 server MUST NOT return NFS4ERR_WRONGSEC to a put filehandle operation whenever it is immediately followed by SECINFO or SECINFO_NO_NAME. The NFSv4.1 server MUST NOT return NFS4ERR_WRONGSEC from SECINFO or SECINFO_NO_NAME. 2.6.3.1.6. Put Filehandle Operation + Nothing The NFSv4.1 server MUST NOT return NFS4ERR_WRONGSEC. 2.6.3.1.7. Put Filehandle Operation + Anything Else "Anything Else" includes OPEN by filehandle. The security policy enforcement applies to the filehandle specified in the put filehandle operation. Therefore PUTFH must return NFS4ERR_WRONGSEC in case of security tuple on the part of the mismatch. This avoids the complexity adding NFS4ERR_WRONGSEC as an allowable error to every other operation. A COMPOUND containing the series put filehandle operation + SECINFO_NO_NAME (style SECINFO_STYLE4_CURRENT_FH) is an efficient way for the client to recover from NFS4ERR_WRONGSEC. The NFSv4.1 server MUST not return NFS4ERR_WRONGSEC to any operation other than a put filehandle operation, LOOKUP, LOOKUPP, and OPEN (by component name). Shepler, et al. Expires September 5, 2007 [Page 31] Internet-Draft NFSv4 Minor Version 1 March 2007 2.7. Minor Versioning To address the requirement of an NFS protocol that can evolve as the need arises, the NFS version 4 protocol contains the rules and framework to allow for future minor changes or versioning. The base assumption with respect to minor versioning is that any future accepted minor version must follow the IETF process and be documented in a standards track RFC. Therefore, each minor version number will correspond to an RFC. Minor version zero of the NFS version 4 protocol is represented by [2], and minor version one is represented by this document [[Comment.2: change "document" to "RFC" when we publish]] . The COMPOUND and CB_COMPOUND procedures support the encoding of the minor version being requested by the client. The following items represent the basic rules for the development of minor versions. Note that a future minor version may decide to modify or add to the following rules as part of the minor version definition. 1. Procedures are not added or deleted To maintain the general RPC model, NFS version 4 minor versions will not add to or delete procedures from the NFS program. 2. Minor versions may add operations to the COMPOUND and CB_COMPOUND procedures. The addition of operations to the COMPOUND and CB_COMPOUND procedures does not affect the RPC model. * Minor versions may append attributes to GETATTR4args, bitmap4, and GETATTR4res. This allows for the expansion of the attribute model to allow for future growth or adaptation. * Minor version X must append any new attributes after the last documented attribute. Since attribute results are specified as an opaque array of per-attribute XDR encoded results, the complexity of adding new attributes in the midst of the current definitions would be too burdensome. 3. Minor versions must not modify the structure of an existing operation's arguments or results. Shepler, et al. Expires September 5, 2007 [Page 32] Internet-Draft NFSv4 Minor Version 1 March 2007 Again the complexity of handling multiple structure definitions for a single operation is too burdensome. New operations should be added instead of modifying existing structures for a minor version. This rule does not preclude the following adaptations in a minor version. * adding bits to flag fields such as new attributes to GETATTR's bitmap4 data type * adding bits to existing attributes like ACLs that have flag words * extending enumerated types (including NFS4ERR_*) with new values 4. Minor versions may not modify the structure of existing attributes. 5. Minor versions may not delete operations. This prevents the potential reuse of a particular operation "slot" in a future minor version. 6. Minor versions may not delete attributes. 7. Minor versions may not delete flag bits or enumeration values. 8. Minor versions may declare an operation as mandatory to NOT implement. Specifying an operation as "mandatory to not implement" is equivalent to obsoleting an operation. For the client, it means that the operation should not be sent to the server. For the server, an NFS error can be returned as opposed to "dropping" the request as an XDR decode error. This approach allows for the obsolescence of an operation while maintaining its structure so that a future minor version can reintroduce the operation. 1. Minor versions may declare attributes mandatory to NOT implement. 2. Minor versions may declare flag bits or enumeration values as mandatory to NOT implement. Shepler, et al. Expires September 5, 2007 [Page 33] Internet-Draft NFSv4 Minor Version 1 March 2007 9. Minor versions may downgrade features from mandatory to recommended, or recommended to optional. 10. Minor versions may upgrade features from optional to recommended or recommended to mandatory. 11. A client and server that supports minor version X should support minor versions 0 (zero) through X-1 as well. 12. Except for infrastructural changes, no new features may be introduced as mandatory in a minor version. This rule allows for the introduction of new functionality and forces the use of implementation experience before designating a feature as mandatory. On the other hand, some classes of features are infrastructural and have broad effects. Allowing such features to not be mandatory complicates implementation of the minor version. 13. A client MUST NOT attempt to use a stateid, filehandle, or similar returned object from the COMPOUND procedure with minor version X for another COMPOUND procedure with minor version Y, where X != Y. 2.8. Non-RPC-based Security Services As described in Section 2.2.1.1.1.1, NFSv4 relies on RPC for identification, authentication, integrity, and privacy. NFSv4 itself provides additional security services as described in the next several subsections. 2.8.1. Authorization Authorization to access a file object via an NFSv4 operation is ultimately determined by the NFSv4 server. A client can predetermine its access to a file object via the OPEN (Section 17.16) and the ACCESS (Section 17.1) operations. Principals with appropriate access rights can modify the authorization on a file object via the SETATTR (Section 17.30) operation. Four attributes that affect access rights are: mode, owner, owner_group, and acl. See Section 5. 2.8.2. Auditing NFSv4 provides auditing on a per file object basis, via the ACL attribute as described in Section 6. It is outside the scope of this specification to specify audit log formats or management policies. Shepler, et al. Expires September 5, 2007 [Page 34] Internet-Draft NFSv4 Minor Version 1 March 2007 2.8.3. Intrusion Detection NFSv4 provides alarm control on a per file object basis, via the ACL attribute as described in Section 6. Alarms may serve as the basis for intrusion detection. It is outside the scope of this specification to specify heuristics for detecting intrusion via alarms. 2.9. Transport Layers 2.9.1. Required and Recommended Properties of Transports NFSv4 works over RDMA and non-RDMA_based transports with the following attributes: o The transport supports reliable delivery of data, which NFSv4 requires but neither NFSv4 nor RPC has facilities for ensuring. [20] o The transport delivers data in the order it was sent. Ordered delivery simplifies detection of transmit errors, and simplifies the sending of arbitrary sized requests and responses, via the record marking protocol [4]. Where an NFS version 4 implementation supports operation over the IP network protocol, any transport used between NFS and IP MUST be among the IETF-approved congestion control transport protocols. At the time this document was written, the only two transports that had the above attributes were TCP and SCTP. To enhance the possibilities for interoperability, an NFS version 4 implementation MUST support operation over the TCP transport protocol. Even if NFS version 4 is used over a non-IP network protocol, it is RECOMMENDED that the transport support congestion control. It is permissible for a connectionless transport to be used under NFSv4.1, however reliable and in-order delivery of data by the connectionless transport is still required. NFSv4.1 assumes that a client transport address and server transport address used to send data over a transport together constitute a connection, even if the underlying transport eschews the concept of a connection. 2.9.2. Client and Server Transport Behavior If a connection-oriented transport (e.g. TCP) is used the client and server SHOULD use long lived connections for at least three reasons: Shepler, et al. Expires September 5, 2007 [Page 35] Internet-Draft NFSv4 Minor Version 1 March 2007 1. This will prevent the weakening of the transport's congestion control mechanisms via short lived connections. 2. This will improve performance for the WAN environment by eliminating the need for connection setup handshakes. 3. The NFSv4.1 callback model differs from NFSv4.0, and requires the client and server to maintain a client-created channel (see Section 2.10.3.4for the server to use. In order to reduce congestion, if a connection-oriented transport is used, and the request is not the NULL procedure, o A requester MUST NOT retry a request unless the connection the request was issued over was disconnected before the reply was received. o A replier MUST NOT silently drop a request, even if the request is a retry. (The silent drop behavior of RPCSEC_GSS [5] does not apply because this behavior happens at the RPCSEC_GSS layer, a lower layer in the request processing). Instead, the replier SHOULD return an appropriate error (see Section 2.10.4.1) or it MAY disconnect the connection. When using RDMA transports there are other reasons for not tolerating retries over the same connection: o RDMA transports use "credits" to enforce flow control, where a credit is a right to a peer to transmit a message. If one peer were to retransmit a request (or reply), it would consume an additional credit. If the replier retransmitted a reply, it would certainly result in an RDMA connection loss, since the requester would typically only post a single receive buffer for each request. If the requester retransmitted a request, the additional credit consumed on the server might lead to RDMA connection failure unless the client accounted for it and decreased its available credit, leading to wasted resources. o RDMA credits present a new issue to the reply cache in NFSv4.1. The reply cache may be used when a connection within a session is lost, such as after the client reconnects. Credit information is a dynamic property of the RDMA connection, and stale values must not be replayed from the cache. This implies that the reply cache contents must not be blindly used when replies are issued from it, and credit information appropriate to the channel must be refreshed by the RPC layer. In addition, the NFSv4.1 requester is not allowed to stop waiting for Shepler, et al. Expires September 5, 2007 [Page 36] Internet-Draft NFSv4 Minor Version 1 March 2007 a reply, as described in Section 2.10.4.2. 2.9.3. Ports Historically, NFS version 2 and version 3 servers have resided on port 2049. The registered port 2049 RFC3232 [21] for the NFS protocol should be the default configuration. NFSv4 clients SHOULD NOT use the RPC binding protocols as described in RFC1833 [22]. 2.10. Session 2.10.1. Motivation and Overview Previous versions and minor versions of NFS have suffered from the following: o Lack of support for exactly once semantics (EOS). This includes lack of support for EOS through server failure and recovery. o Limited callback support, including no support for sending callbacks through firewalls, and races between responses from normal requests, and callbacks. o Limited trunking over multiple network paths. o Requiring machine credentials for fully secure operation. Through the introduction of a session, NFSv4.1 addresses the above shortfalls with practical solutions: o EOS is enabled by a reply cache with a bounded size, making it feasible to keep on persistent storage and enable EOS through server failure and recovery. One reason that previous revisions of NFS did not support EOS was because some EOS approaches often limited parallelism. As will be explained in Section 2.10.4), NFSv4.1 supports both EOS and unlimited parallelism. o The NFSv4.1 client provides creates transport connections and gives them to the server for sending callbacks, thus solving the firewall issue (Section 17.34). Races between responses from client requests, and callbacks caused by the requests are detected via the session's sequencing properties which are a byproduct of EOS (Section 2.10.4.3). o The NFSv4.1 client can add an arbitrary number of connections to the session, and thus provide trunking (Section 2.10.3.4.1). Shepler, et al. Expires September 5, 2007 [Page 37] Internet-Draft NFSv4 Minor Version 1 March 2007 o The NFSv4.1 session produces a session key independent of client and server machine credentials which can be used to compute a digest for protecting key session management operations Section 2.10.6.3). o The NFSv4.1 client can also create secure RPCSEC_GSS contexts for use by the session's callback channel that do not require the server to authenticate to a client machine principal (Section 2.10.6.2). A session is a dynamically created, long-lived server object created by a client, used over time from one or more transport connections. Its function is to maintain the server's state relative to the connection(s) belonging to a client instance. This state is entirely independent of the connection itself, and indeed the state exists whether the connection exists or not (though locks, delegations, etc. and generally expire in the extended absence of an open connection). The session in effect becomes the object representing an active client on a set of zero or more connections. 2.10.2. NFSv4 Integration Sessions are part of NFSv4.1 and not NFSv4.0. Normally, a major infrastructure change like sessions would require a new major version number to an RPC program like NFS. However, because NFSv4 encapsulates its functionality in a single procedure, COMPOUND, and because COMPOUND can support an arbitrary number of operations, sessions are almost trivially added. COMPOUND includes a minor version number field, and for NFSv4.1 this minor version is set to 1. When the NFSv4 server processes a COMPOUND with the minor version set to 1, it expects a different set of operations than it does for NFSv4.0. One operation it expects is the SEQUENCE operation, which is required for every COMPOUND that operates over an established session. 2.10.2.1. SEQUENCE and CB_SEQUENCE In NFSv4.1, when the SEQUENCE operation is present, it is always the first operation in the COMPOUND procedure. The primary purpose of SEQUENCE is to carry the session identifier. The session identifier associates all other operations in the COMPOUND procedure with a particular session. SEQUENCE also contains required information for maintaining EOS (see Section 2.10.4). Session-enabled NFSv4.1 COMPOUND requests thus have the form: Shepler, et al. Expires September 5, 2007 [Page 38] Internet-Draft NFSv4 Minor Version 1 March 2007 +-----+--------------+-----------+------------+-----------+---- | tag | minorversion | numops |SEQUENCE op | op + args | ... | | (== 1) | (limited) | + args | | +-----+--------------+-----------+------------+-----------+---- and the reply's structure is: +------------+-----+--------+-------------------------------+--// |last status | tag | numres |status + SEQUENCE op + results | // +------------+-----+--------+-------------------------------+--// //-----------------------+---- // status + op + results | ... //-----------------------+---- A CB_COMPOUND procedure request and reply has a similar form, but instead of a SEQUENCE operation, there is a CB_SEQUENCE operation, and there is an additional field called "callback_ident", which is superfluous in NFSv4.1. CB_SEQUENCE has the same information as SEQUENCE, but includes other information needed to solve callback races (Section 2.10.4.3). 2.10.2.2. Client ID and Session Association Sessions are subordinate to the client ID (Section 2.4). Each client ID can have zero or more active sessions. A client ID, and a session bound to it are required to do anything useful in NFSv4.1. Each time a session is used, the state leased to its associated client ID is automatically renewed. State such as share reservations, locks, delegations, and layouts (Section 1.4.4) is tied to the client ID, not the sessions of the client ID. Successive state changing operations from a given state owner can go over different sessions, as long each session is associated with the same client ID. Callbacks can arrive over a different session than the session that sent the operation the acquired the state that the callback is for. For example, if session A is used to acquire a delegation, a request to recall the delegation can arrive over session B. 2.10.3. Channels Each session has one or two channels: the "operation" or "fore" channel used for ordinary requests from client to server, and the "back" channel, used for callback requests from server to client. The session allocates resources for each channel, including separate reply caches (see Section 2.10.4.1). These resources are for the most part specified at time the session is created. Shepler, et al. Expires September 5, 2007 [Page 39] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.3.1. Operation Channel The operation channel carries COMPOUND requests and responses. A session always has an operation channel. 2.10.3.2. Backchannel The backchannel carries CB_COMPOUND requests and responses. Whether there is a backchannel or not is a decision of the client; NFSv4.1 servers MUST support backchannels. 2.10.3.3. Session and Channel Association Because there are at most two channels per session, and because each channel has a distinct purpose, channels are not assigned identifiers. The operation and backchannel are implicitly created and associated when the session is created. 2.10.3.4. Connection and Channel Association Each channel is associated with zero or more transport connections. A connection can be bound to one channel or both channels of a session; the client and server negotiate whether a connection will carry traffic for one channel or both channels via the CREATE_SESSION (Section 17.36) and the BIND_CONN_TO_SESSION (Section 17.34) operations. When a session is created via CREATE_SESSION, it is automatically bound to the operation channel, and optionally the backchannel. If the client does not specify connecting binding enforcement when the session is created, then additional connections are automatically bound to the operation channel when the are used with a SEQUENCE operation that has the session's sessionid. A connection MAY be bound to the channels of other sessions. The client decides, and the NFSv4.1 server MUST allow it. A connection MAY be bound to the channels of other sessions of other clientids. Again, the client decides, and the server MUST allow it. It is permissible for connections of multiple types to be bound to the same channel. For example a TCP and RDMA connection can be bound to the operation channel. In the event an RDMA and non-RDMA connection are bound to the same channel, the maximum number of slots must be at least one more than the total number of credits. This way if all RDMA credits are use, the non-RDMA connection can have at least one outstanding request. It is permissible for a connection of one type to be bound to the operation channel, and another type bound to the backchannel. Shepler, et al. Expires September 5, 2007 [Page 40] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.3.4.1. Trunking A client is allowed to issue EXCHANGE_ID multiple times to the same server. The client may be unaware that two different server network addresses refer to the same server. The use of EXCHANGE_ID allows a client to become aware that an additional network address refers to a server the client already has an established client ID and session for. The eir_server_owner and eir_server_scope results from EXCHANGE_ID give a client a hint that the server it is connected to may be the same as the server it is connected to via another connection. When EXCHANGE_ID is issued over two different connections, and each return the same eir_server_owner.so_major_id and eir_server_scope, the client treats the connections as connected to the same server (subject to verification, as described later in this section (Paragraph 2), even if the destination network addresses are different). As long two unrelated servers have not selected and returned a conflicting pair of eir_major_id and eir_server_scope, or unless the client has used different co_ownerid values in each EXCHANGE_ID request, or the server has lost client ID state (e.g. the server has rebooted) the server MUST return the same eir_clientid result. Otherwise, the client and server use the common eir_clientid to identify the client. The eir_server_owner.so_minor_id field allows the server to control binding of connections to sessions. When two connections have a matching eir_server_scope, so_major_id and so_minor_id, the client may bind both connections to a common session; this is session trunking. When two connections have a matching so_major_id and eir_server_scope, but different so_minor_id, the client will need to create a new session for the client ID in order to use the connection; this is client ID trunking. In either session or client ID trunking, the bandwidth capacity can scale with the number of connections. When two servers over two connections claim matching or partially matching eir_server_owner, eir_server_scope, and eir_clientid values the client does not have to trust the servers' claims. The client may verify these claims before trunking traffic in the following ways: o For session trunking, clients and servers can reliably verify if connections between different network paths are in fact bound to the same NFSv4.1 server and usable on the same session. The SET_SSV (Section 17.47) operation allows a client and server to establish a unique, shared key value (the SSV). When a new connection is bound to the session (via the BIND_CONN_TO_SESSION operation, see Section 17.34), the client offers a digest that is based on the SSV. If the client mistakenly tries to bind a connection to a session of a wrong server, the server will either reject the attempt because it is not aware of the session Shepler, et al. Expires September 5, 2007 [Page 41] Internet-Draft NFSv4 Minor Version 1 March 2007 identifier of the BIND_CONN_TO_SESSION arguments, or it will reject the attempt because the digest for the SSV does not match what the server expects. Even if the server mistakenly or maliciously accepts the connection bind attempt, the digest it computes in the response will not be verified by the client, the client will know it cannot use the connection for trunking the specified channel. o In the case of client ID trunking, the client can use RPCSEC_GSS to verify that each connection is aimed at the same server. When the client invokes EXCHANGE_ID, it should use RPCSEC_GSS. If each RPCSEC_GSS context over each connection has the same server principal, then -- barring a compromise of the server's GSS credentials -- the servers at the end of each connection are the same. 2.10.4. Exactly Once Semantics Via the session, NFSv4.1 offers exactly once semantics (EOS) for requests sent over a channel. EOS is supported on both the operation and back channels. Each COMPOUND or CB_COMPOUND request that is issued with a leading SEQUENCE or CB_SEQUENCE operation MUST be executed by the receiver exactly once. This requirement is regardless whether the request is issued with reply caching specified (see Section 2.10.4.1.2). The requirement holds even if the requester is issuing the request over a session created between a pNFS data client and pNFS data server. The rationale for this requirement is understood by categorizing requests into three classifications: o Nonidempotent requests. o Idempotent modifying requests. o Idempotent non-modifying requests. An example of a non-idempotent request is RENAME. If is obvious that if a replier executes the same RENAME request twice, and the first execution succeeds, the re-execution will fail. If the replier returns the result from the re-execution, this result is incorrect. Therefore, EOS is required for nonidempotent requests. An example of an idempotent modifying request is a COMPOUND request containing a WRITE operation. Repeated execution of the same WRITE has the same effect as execution of that write once. Nevertheless, putting enforcing EOS for WRITEs and other idempotent modifying requests is necessary to avoid data corruption. Shepler, et al. Expires September 5, 2007 [Page 42] Internet-Draft NFSv4 Minor Version 1 March 2007 Suppose a client issues WRITEs A, B, C to a noncompliant server that does not enforce EOS, and receives no response, perhaps due to a network partition. The client reconnects to the server and re-issues all three WRITEs. Now, the server has outstanding two instances of each of A, B, and C. The server can be in a situation in which it executes and replies to the retries of A, B, and C while the first A, B, and C are still waiting around in the server's I/O system for some resource. Upon receiving the replies to the second attempts of WRITEs A, B, and C, the client believes its writes are done so it is free to do issue WRITE D which overlaps the range of one or more of A, B, C. If any of A, B, or C are subsequently are executed for the second time, then what has been written by D can be overwritten and thus corrupted. Note that it is not required the server cache the reply to the modifying operation to avoid data corruption (but if the client specified the reply to be cached, the server must cache it). An example of an idempotent non-modifying request is a COMPOUND containing SEQUENCE, PUTFH, READLINK and nothing else. The re- execution of a such a request will not cause data corruption, or produce an incorrect result. Nonetheless, for simplicity, the replier MUST enforce EOS for such requests. 2.10.4.1. Slot Identifiers and Reply Cache The RPC layer provides a transaction ID (xid), which, while required to be unique, is not especially convenient for tracking requests. The xid is only meaningful to the requester it cannot be interpreted at the replier except to test for equality with previously issued requests. Because RPC operations may be completed by the replier in any order, many transaction IDs may be outstanding at any time. The requester may therefore perform a computationally expensive lookup operation in the process of demultiplexing each reply. In the NFSv4.1, there is a limit to the number of active requests. This immediately enables a computationally efficient index for each request which is designated as a Slot Identifier, or slotid. When the requester issues a new request, it selects a slotid in the range 0..N-1, where N is the replier's current "outstanding requests" limit granted to the requester on the session over which the request is to be issued. The value of N outstanding requests starts out as the value of ca_maxrequests (Section 17.36), but can be adjusted by the response to SEQUENCE or CB_SEQUENCE as described later in this section. The slotid must be unused by any of the requests which the requester has already active on the session. "Unused" here means the requester has no outstanding request for that slotid. Because the Shepler, et al. Expires September 5, 2007 [Page 43] Internet-Draft NFSv4 Minor Version 1 March 2007 slot id is always an integer in the range 0..N-1, requester implementations can use the slotid from a replier response to efficiently match responses with outstanding requests, such as, for example, by using the slotid to index into an outstanding request array. This can be used to avoid expensive hashing and lookup functions in the performance-critical receive path. The sequenceid, which accompanies the slotid in each request, is for an important check at the server: it must be able to be determined efficiently whether a request using a certain slotid is a retransmit or a new, never-before-seen request. It is not feasible for the client to assert that it is retransmitting to implement this, because for any given request the client cannot know the server has seen it unless the server actually replies. Of course, if the client has seen the server's reply, the client would not retransmit. The sequenceid MUST increase monotonically for each new transmit of a given slotid, and MUST remain unchanged for any retransmission. The server must in turn compare each newly received request's sequenceid with the last one previously received for that slotid, to see if the new request is: o A new request, in which the sequenceid is one greater than that previously seen in the slot (accounting for sequence wraparound). The replier proceeds to execute the new request. o A retransmitted request, in which the sequenceid is equal to that last seen in the slot. Note that this request may be either complete, or in progress. The replier performs replay processing in these cases. o A misordered replay, in which the sequenceid is less than (accounting for sequence wraparound) than that previously seen in the slot. The replier MUST return NFS4ERR_SEQ_MISORDERED (as the result from SEQUENCE or CB_SEQUENCE). o A misordered new request, in which the sequenceid is two or more than (accounting for sequence wraparound) than that previously seen in the slot. Note that because the sequenceid must wraparound one it reaches 0xFFFFFFFF, a misordered new request and a misordered replay cannot be distinguished. Thus, the replier MUST return NFS4ERR_SEQ_MISORDERED (as the result from SEQUENCE or CB_SEQUENCE). Unlike the XID, the slotid is always within a specific range; this has two implications. The first implication is that for a given session, the replier need only cache the results of a limited number of COMPOUND requests. The second implication derives from the first, Shepler, et al. Expires September 5, 2007 [Page 44] Internet-Draft NFSv4 Minor Version 1 March 2007 which is unlike XID-indexed reply caches (also know as duplicate request caches - DRCs), the slotid-based reply cache cannot be overflowed. Through use of the sequenceid to identify retransmitted requests, the replier does not need to actually cache the request itself, reducing the storage requirements of the reply cache further. These new facilities makes it practical to maintain all the required entries for an effective reply cache. The slotid and sequenceid therefore take over the traditional role of the XID and port number in the replier reply cache implementation, and the session replaces the IP address. This approach is considerably more portable and completely robust - it is not subject to the frequent reassignment of ports as clients reconnect over IP networks. In addition, the RPC XID is not used in the reply cache, enhancing robustness of the cache in the face of any rapid reuse of XIDs by the client. [[Comment.3: We need to discuss the requirements of the client for changing the XID.]] The slotid information is included in each request, without violating the minor versioning rules of the NFSv4.0 specification, by encoding it in the SEQUENCE operation within each NFSv4.1 COMPOUND and CB_COMPOUND procedure. The operation easily piggybacks within existing messages. [[Comment.4: Need a better term than piggyback]] The receipt of a new sequenced request arriving on any valid slot is an indication that the previous reply cache contents of that slot may be discarded. The SEQUENCE (and CB_SEQUENCE) operation also carries a "highest_slotid" value which carries additional client slot usage information. The requester must always provide a slotid representing the outstanding request with the highest-numbered slot value. The requester should in all cases provide the most conservative value possible, although it can be increased somewhat above the actual instantaneous usage to maintain some minimum or optimal level. This provides a way for the requester to yield unused request slots back to the replier, which in turn can use the information to reallocate resources. The replier responds with both a new target highest_slotid, and an enforced highest_slotid, described as follows: o The target highest_slotid is an indication to the requester of the highest_slotid the replier wishes the requester to be using. This permits the replier to withdraw (or add) resources from a requester that has been found to not be using them, in order to more fairly share resources among a varying level of demand from other requesters. The requester must always comply with the Shepler, et al. Expires September 5, 2007 [Page 45] Internet-Draft NFSv4 Minor Version 1 March 2007 replier's value updates, since they indicate newly established hard limits on the requester's access to session resources. However, because of request pipelining, the requester may have active requests in flight reflecting prior values, therefore the replier must not immediately require the requester to comply. o The enforced highest_slotid indicates the highest slotid the requester is permitted to use on a subsequent SEQUENCE or CB_SEQUENCE operation. The requester is required to use the lowest available slot when issuing a new request. This way, the replier may be able to retire slot entries faster. However, where the replier is actively adjusting its granted maximum request count (i.e. the highest_slotid) to the requester, it will not not be able to use just the receipt of the slotid and highest_slotid in the request. Neither the slotid nor the highest_slotid used in a request may reflect the replier's current idea of the requester's session limit, because the request may have been sent from the requester before the update was received. Therefore, in the downward adjustment case, the replier may have to retain a number of reply cache entries at least as large as the old value of maximum requests outstanding, until operation sequencing rules allow it to infer that the requester has seen its reply. 2.10.4.1.1. Errors from SEQUENCE and CB_SEQUENCE Any time SEQUENCE or CB_SEQUENCE return an error, the sequenceid of the slot MUST NOT change. The replier MUST NOT modify the reply cache entry for the slot whenever an error is returned from SEQUENCE or CB_SEQUENCE. 2.10.4.1.2. Optional Reply Caching On a per-request basis the requester can choose to direct the replier to cache the reply to all operations after the first operation (SEQUENCE or CB_SEQUENCE) via the sa_cachethis or csa_cachethis fields of the arguments to SEQUENCE or CB_SEQUENCE. The reason it would not direct the replier to cache the entire reply is that the request is composed of all idempotent operations [20]. Caching the reply may offer little benefit, and if the reply is too large (see Section 2.10.4.4), it may not be cacheable anyway. Whether the requester requests the reply to be cached or not has no effect on the slot processing. If the results of SEQUENCE or CB_SEQUENCE are NFS4_OK, then the slot's sequenceid MUST be incremented by one. If a requester does not direct the replier to cache, the reply, the replier MUST do one of following: Shepler, et al. Expires September 5, 2007 [Page 46] Internet-Draft NFSv4 Minor Version 1 March 2007 o The replier can cache the entire original reply. Even though sa_cachethis or csa_cachethis are FALSE, the replier is always free to cache. It may choose this approach in order to simplify implementation. o The replier enters into its reply cache a reply consisting of the original results to the SEQUENCE or CB_SEQUENCE operation, followed by the error NFS4ERR_RETRY_UNCACHED_REP. Thus if the requester later retries the request, it will get NFS4ERR_RETRY_UNCACHE_REP. 2.10.4.1.3. Multiple Connections and Sharing the Reply Cache Multiple connections can be bound to a session's channel, hence the connections share the same table of slotids. For connections over non-RDMA transports like TCP, there are no particular considerations. Considerations for multiple RDMA connections sharing a slot table are discussed in Section 2.10.5.1. [[Comment.5: Also need to discuss when RDMA and non-RDMA share a slot table.]] 2.10.4.2. Retry and Replay A client MUST NOT retry a request, unless the connection it used to send the request disconnects. The client can then reconnect and resend the request, or it can resend the request over a different connection. In the case of the server resending over the backchannel, it cannot reconnect, and either resends the request over another connection that the client has bound to the backchannel, or if there is no other backchannel connection, waits for the client to bind a connection to the backchannel. A client MUST wait for a reply to a request before using the slot for another request. If it does not wait for a reply, then the client does not know what sequenceid to use for the slot on its next request. For example, suppose a client sends a request with sequenceid 1, and does not wait for the response. The next time it uses the slot, it sends the new request with sequenceid 2. If the server has not seen the request with sequenceid 1, then the server is expecting sequenceid 2, and rejects the client's new request with NFS4ERR_SEQ_MISORDERED (as the result from SEQUENCE or CB_SEQUENCE). RDMA fabrics do not guarantee that the memory handles (Steering Tags) within each RDMA three-tuple are valid on a scope [[Comment.6: What is a three-tuple?]] outside that of a single connection. Therefore, handles used by the direct operations become invalid after connection loss. The server must ensure that any RDMA operations which must be replayed from the reply cache use the newly provided handle(s) from the most recent request. Shepler, et al. Expires September 5, 2007 [Page 47] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.4.3. Resolving server callback races with sessions It is possible for server callbacks to arrive at the client before the reply from related forward channel operations. For example, a client may have been granted a delegation to a file it has opened, but the reply to the OPEN (informing the client of the granting of the delegation) may be delayed in the network. If a conflicting operation arrives at the server, it will recall the delegation using the callback channel, which may be on a different transport connection, perhaps even a different network. In NFSv4.0, if the callback request arrives before the related reply, the client may reply to the server with an error. The presence of a session between client and server alleviates this issue. When a session is in place, each client request is uniquely identified by its { slotid, sequenceid } pair. By the rules under which slot entries (reply cache entries) are retired, the server has knowledge whether the client has "seen" each of the server's replies. The server can therefore provide sufficient information to the client to allow it to disambiguate between an erroneous or conflicting callback and a race condition. For each client operation which might result in some sort of server callback, the server should "remember" the { slotid, sequenceid } pair of the client request until the slotid retirement rules allow the server to determine that the client has, in fact, seen the server's reply. Until the time the { slotid, sequenceid } request pair can be retired, any recalls of the associated object MUST carry an array of these referring identifiers (in the CB_SEQUENCE operation's arguments), for the benefit of the client. After this time, it is not necessary for the server to provide this information in related callbacks, since it is certain that a race condition can no longer occur. The CB_SEQUENCE operation which begins each server callback carries a list of "referring" { slotid, sequenceid } tuples. If the client finds the request corresponding to the referring slotid and sequenced id be currently outstanding (i.e. the server's reply has not been seen by the client), it can determine that the callback has raced the reply, and act accordingly. The client must not simply wait forever for the expected server reply to arrive on any of the session's operations channels, because it is possible that they will be delayed indefinitely. However, it should wait for a period of time, and if the time expires it can provide a more meaningful error such as NFS4ERR_DELAY. [[Comment.7: We need to consider the clients' options here, and Shepler, et al. Expires September 5, 2007 [Page 48] Internet-Draft NFSv4 Minor Version 1 March 2007 describe them... NFS4ERR_DELAY has been discussed as a legal reply to CB_RECALL?]] There are other scenarios under which callbacks may race replies, among them pNFS layout recalls, described in Section 12.5.4.2 [[Comment.8: fill in the blanks w/others, etc...]] 2.10.4.4. COMPOUND and CB_COMPOUND Construction Issues Very large requests and replies may pose both buffer management issues (especially with RDMA) and reply cache issues. When the session is created, (Section 17.36) the client and server negotiate the maximum sized request they will send or process (ca_maxrequestsize), the maximum sized reply they will return or process (ca_maxresponsesize), and the maximum sized reply they will store in the reply cache (ca_maxresponsesize_cached). If a request exceeds ca_maxrequestsize, the reply will have the status NFS4ERR_REQ_TOO_BIG. A replier may return NFS4ERR_REQ_TOO_BIG as the status for first operation (SEQUENCE or CB_SEQUENCE) in the request, or it may chose to return it on a subsequent operation. If a reply exceeds ca_maxresponsesize, the reply will have the status NFS4ERR_REP_TOO_BIG. A replier may return NFS4ERR_REP_TOO_BIG as the status for first operation (SEQUENCE or CB_SEQUENCE) in the request, or it may chose to return it on a subsequent operation. If sa_cachethis or csa_cachethis are TRUE, then the replier MUST cache a reply except if an error is returned by the SEQUENCE or CB_SEQUENCE operation (see Section 2.10.4.1.1). If the reply exceeds ca_maxresponsesize_cached, (and sa_cachethis or csa_cachethis are TRUE) then the server MUST return NFS4ERR_REP_TOO_BIG_TO_CACHE. Even if NFS4ERR_REP_TOO_BIG_TO_CACHE (or any other error for that matter) is returned on a operation other than first operation (SEQUENCE or CB_SEQUENCE), then the reply MUST be cached if sa_cachethis or csa_cachethis are TRUE. For example, if a COMPOUND has eleven operations, including SEQUENCE, the fifth operation is a RENAME, and the tenth operation is a READ for one million bytes, server may return NFS4ERR_REP_TOO_BIG_TO_CACHE on the tenth operation. Since the server executed several operations, especially the non-idempotent RENAME, the client's request to cache the reply needs to be honored in order for correct operation of exactly once semantics. If the client retries the request, the server will have cached a reply that contains results for ten of the eleven requested operations, with the tenth operation having a status of NFS4ERR_REP_TOO_BIG_TO_CACHE. A client needs to take care that when sending operations that change the current filehandle (except for PUTFH, PUTPUBFH, and PUTROOTFH) Shepler, et al. Expires September 5, 2007 [Page 49] Internet-Draft NFSv4 Minor Version 1 March 2007 that it not exceed the maximum reply buffer before the GETFH operation. Otherwise the client will have to retry the operation that changed the current filehandle, in order obtain the desired filehandle. For the OPEN operation (see Section 17.16), retry is not always available as an option. The following guidelines for the handling of filehandle changing operations are advised: o A client SHOULD issue GETFH immediately after a current filehandle changing operation. This is especially important after any current filehandle changing non-idempotent operation. It is critical to issue GETFH immediately after OPEN. o A server MAY return NFS4ERR_REP_TOO_BIG or NFS4ERR_REP_TOO_BIG_TO_CACHE (if sa_cachethis is TRUE) on a filehandle changing operation if the reply would be too large on the next operation. o A server SHOULD return NFS4ERR_REP_TOO_BIG or NFS4ERR_REP_TOO_BIG_TO_CACHE (if sa_cachethis is TRUE) on a filehandle changing non-idempotent operation if the reply would be too large on the next operation, especially if the operation is OPEN. o A server MAY return NFS4ERR_UNSAFE_COMPOUND if it looks at the next operation after a non-idempotent current filehandle changing operation, and finds it is not GETFH. The server would do this if it is unable to determine in advance whether the total response size would exceed ca_maxresponsesize_cached or ca_maxresponsesize. 2.10.4.5. Persistence Since the reply cache is bounded, it is practical for the server reply cache to persist across server reboots, and to be kept in stable storage (a client's reply cache for callbacks need not persist across client reboots unless the client intends for its session and other state to persist across reboots). o The slot table including the sequenceid and cached reply for each slot. o The sessionid. o The client ID. o The SSV (see Section 2.10.6.3). The CREATE_SESSION (see Section 17.36 operation determines the persistence of the reply cache. Shepler, et al. Expires September 5, 2007 [Page 50] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.5. RDMA Considerations A complete discussion of the operation of RPC-based protocols atop RDMA transports is in [RPCRDMA]. A discussion of the operation of NFSv4, including NFSv4.1 over RDMA is in [NFSDDP]. Where RDMA is considered, this specification assumes the use of such a layering; it addresses only the upper layer issues relevant to making best use of RPC/RDMA. 2.10.5.1. RDMA Connection Resources RDMA requires its consumers to register memory and post buffers of a specific size and number for receive operations. Registration of memory can be a relatively high-overhead operation, since it requires pinning of buffers, assignment of attributes (e.g. readable/writable), and initialization of hardware translation. Preregistration is desirable to reduce overhead. These registrations are specific to hardware interfaces and even to RDMA connection endpoints, therefore negotiation of their limits is desirable to manage resources effectively. Following the basic registration, these buffers must be posted by the RPC layer to handle receives. These buffers remain in use by the RPC/NFSv4 implementation; the size and number of them must be known to the remote peer in order to avoid RDMA errors which would cause a fatal error on the RDMA connection. NFSv4.1 manages slots as resources on a per session basis (see Section 2.10), while RDMA connections manage credits on a per connection basis. This means that in order for a peer to send data over RDMA to a remote buffer, it has to have both an NFSv4.1 slot, and an RDMA credit. 2.10.5.2. Flow Control NFSv4.0 and all previous versions do not provide for any form of flow control; instead they rely on the windowing provided by transports like TCP to throttle requests. This does not work with RDMA, which provides no operation flow control and will terminate a connection in error when limits are exceeded. Limits such as maximum number of requests outstanding are therefore negotiated when a session is created (see the ca_maxrequests field in Section 17.36). These limits then provide the maxima each session's channels' connections must operate within. RDMA connections are managed within these limits as described in section 3.3 of [RPCRDMA]; if there are multiple RDMA connections, then the maximum requests for a channel will be divided among the RDMA connections. The limits may also be Shepler, et al. Expires September 5, 2007 [Page 51] Internet-Draft NFSv4 Minor Version 1 March 2007 modified dynamically at the server's choosing by manipulating certain parameters present in each NFSv4.1 request. In addition, the CB_RECALL_SLOT callback operation (see Section 19.8 can be issued by a server to a client to return RDMA credits to the server, thereby lowering the maximum number of requests a client can have outstanding to the server. 2.10.5.3. Padding Header padding is requested by each peer at session initiation (see the csa_headerpadsize argument to CREATE_SESSION in Section 17.36), and subsequently used by the RPC RDMA layer, as described in [RPCRDMA]. Zero padding is permitted. Padding leverages the useful property that RDMA receives preserve alignment of data, even when they are placed into anonymous (untagged) buffers. If requested, client inline writes will insert appropriate pad bytes within the request header to align the data payload on the specified boundary. The client is encouraged to add sufficient padding (up to the negotiated size) so that the "data" field of the NFSv4.1 WRITE operation is aligned. Most servers can make good use of such padding, which allows them to chain receive buffers in such a way that any data carried by client requests will be placed into appropriate buffers at the server, ready for file system processing. The receiver's RPC layer encounters no overhead from skipping over pad bytes, and the RDMA layer's high performance makes the insertion and transmission of padding on the sender a significant optimization. In this way, the need for servers to perform RDMA Read to satisfy all but the largest client writes is obviated. An added benefit is the reduction of message round trips on the network - a potentially good trade, where latency is present. The value to choose for padding is subject to a number of criteria. A primary source of variable-length data in the RPC header is the authentication information, the form of which is client-determined, possibly in response to server specification. The contents of COMPOUNDs, sizes of strings such as those passed to RENAME, etc. all go into the determination of a maximal NFSv4 request size and therefore minimal buffer size. The client must select its offered value carefully, so as not to overburden the server, and vice- versa. The payoff of an appropriate padding value is higher performance. Sender gather: |RPC Request|Pad bytes|Length| -> |User data...| \------+---------------------/ \ \ \ \ Receiver scatter: \-----------+- ... /-----+----------------\ \ \ Shepler, et al. Expires September 5, 2007 [Page 52] Internet-Draft NFSv4 Minor Version 1 March 2007 |RPC Request|Pad|Length| -> |FS buffer|->|FS buffer|->... In the above case, the server may recycle unused buffers to the next posted receive if unused by the actual received request, or may pass the now-complete buffers by reference for normal write processing. For a server which can make use of it, this removes any need for data copies of incoming data, without resorting to complicated end-to-end buffer advertisement and management. This includes most kernel-based and integrated server designs, among many others. The client may perform similar optimizations, if desired. 2.10.5.4. Dual RDMA and Non-RDMA Transports Some RDMA transports (for example see [RDDP]), [[Comment.9: need xref]] require a "streaming" (non-RDMA) phase, where ordinary traffic might flow before "stepping" up to RDMA mode, commencing RDMA traffic. Some RDMA transports start connections always in RDMA mode. NFSv4.1 allows, but does not assume, a streaming phase before RDMA mode. When a connection is bound to a session, the client and server negotiate whether the connection is used in RDMA or non-RDMA mode (see Section 17.36 and Section 17.34). 2.10.6. Sessions Security 2.10.6.1. Session Callback Security Via session connection binding, NFSv4.1 improves security over that provided by NFSv4.0 for the callback channel. The connection is client-initiated (see Section 17.34), and subject to the same firewall and routing checks as the operations channel. The connection cannot be hijacked by an attacker who connects to the client port prior to the intended server. At the client's option (see Section 17.36 binding is fully authenticated before being activated (see Section 17.34). Traffic from the server over the callback channel is authenticated exactly as the client specifies (see Section 2.10.6.2). 2.10.6.2. Backchannel RPC Security When the NFSv4.1 client establishes the backchannel, it informs the server what security flavors and principals it must use when sending requests over the backchannel. If the security flavor is RPCSEC_GSS, the client expresses the principal in the form of an established RPCSEC_GSS context. The server is free to use any flavor/principal combination the server offers, but MUST NOT use unoffered combinations. This way, the client does not have to provide a target GSS principal Shepler, et al. Expires September 5, 2007 [Page 53] Internet-Draft NFSv4 Minor Version 1 March 2007 as it did with NFSv4.0, and the server does not have to implement an RPCSEC_GSS initiator as it did with NFSv4.0. [[Comment.10: xrefs]] The CREATE_SESSION (Section 17.36) and BACKCHANNEL_CTL (Section 17.33) operations allow the client to specify flavor/ principal combinations. 2.10.6.3. Protection from Unauthorized State Changes Under some conditions, NFSv4.0 is vulnerable to a denial of service issue with respect to its state management. The attack works via an unauthorized client faking an open_owner4, an open_owner/lock_owner pair, or stateid, combined with a seqid. The operation is sent to the NFSv4 server. The NFSv4 server accepts the state information, and as long as any status code from the result of this operation is not NFS4ERR_STALE_CLIENTID, NFS4ERR_STALE_STATEID, NFS4ERR_BAD_STATEID, NFS4ERR_BAD_SEQID, NFS4ERR_BADXDR, NFS4ERR_RESOURCE, or NFS4ERR_NOFILEHANDLE, the sequence number is incremented. When the authorized client issues an operation, it gets back NFS4ERR_BAD_SEQID, because its idea of the current sequence number is off by one. The authorized client's recovery options are pretty limited, with SETCLIENTID, followed by complete reclaim of state, which may or may not succeed completely. That qualifies as a denial of service attack. If the client uses RPCSEC_GSS authentication and integrity, and every client maps each open_owner and lock_owner one and only one principal, and the server enforces this binding, then the conditions leading to vulnerability to the denial of service do not exist. One should keep in mind that if AUTH_SYS is being used, far simpler easier denial of service and other attacks are possible. With NFSv4.1 sessions, the per-operation sequence number is ignored (see Section 8.13) therefore the NFSv4.0 denial of service vulnerability described above does not apply. However as described to this point in the specification, an attacker could forge the sessionid and issue a SEQUENCE with a slot id that he expects the legitimate client to use next. The legitimate client could then use the slotid with the same sequence number, and the server returns the attacker's result from the replay cache, thereby disrupting the legitimate client. If we give each NFSv4.1 user their own session, and each user uses RPCSEC_GSS authentication and integrity, then the denial of service issue is solved, at the cost of additional per session state. The alternative NFSv4.1 specifies is described as follows. Shepler, et al. Expires September 5, 2007 [Page 54] Internet-Draft NFSv4 Minor Version 1 March 2007 Transport connections MUST be bound to a session by the client. The server MUST return an error to an operation (other than the operation that binds the connection to the session) that uses an unbound connection. As a simplification, the transport connection used by CREATE_SESSION (see Section 17.36) is automatically bound to the session. Additional connections are bound to a session via BIND_CONN_TO_SESSION (see Section 17.34). To prevent attackers from issuing BIND_CONN_TO_SESSION operations, the arguments to BIND_CONN_TO_SESSION include a digest of a shared secret called the secret session verifier (SSV) that only the client and server know. The digest is created via a one way, collision resistant hash function, making it intractable for the attacker to forge. The SSV is sent to the server via SET_SSV (see Section 17.47). To prevent eavesdropping, a SET_SSV for the SSV SHOULD be protected via RPCSEC_GSS with the privacy service. The SSV can be changed by the client at any time, by any principal. However several aspects of SSV changing prevent an attacker from engaging in a successful denial of service attack: o A SET_SSV on the SSV does not replace the SSV with the argument to SET_SSV. Instead, the current SSV on the server is logically exclusive ORed (XORed) with the argument to SET_SSV. SET_SSV MUST NOT be called with an SSV value that is zero. o The arguments to and results of SET_SSV include digests of the old and new SSV, respectively. o Because the initial value of the SSV is zero, therefore known, the client that opts for connecting binding enforcement, MUST issue at least one SET_SSV operation before the first BIND_CONN_TO_SESSION operation. A client SHOULD issue SET_SSV as soon as a session is created. If a connection is disconnected, BIND_CONN_TO_SESSION is required to bind a connection to the session, even if the connection that was disconnected was the one CREATE_SESSION was created with. If a client is assigned a machine principal then the client SHOULD use the machine principal's RPCSEC_GSS context to privacy protect the SSV from eavesdropping during the SET_SSV operation. If a machine principal is not being used, then the client MAY use the non-machine principal's RPCSEC_GSS context to privacy protect the SSV. The server MUST accept either type of principal. A client SHOULD change the SSV each time a new principal uses the session. Shepler, et al. Expires September 5, 2007 [Page 55] Internet-Draft NFSv4 Minor Version 1 March 2007 Here are the types of attacks that can be attempted by an attacker named Eve, and how the connection to session binding approach addresses each attack: o If the Eve creates a connection after the legitimate client establishes an SSV via privacy protection from a machine principal's RPCSEC_GSS session, she does not know the SSV and so cannot compute a digest that BIND_CONN_TO_SESSION will accept. Users on the legitimate client cannot be disrupted by Eve. o If Eve is the first one log into the legitimate client, and the client does not use machine principals, then Eve can cause an SSV to be created via the legitimate client's NFSv4.1 implementation, protected by the RPCSEC_GSS context created by the legitimate client (which uses Eve's GSS principal and credentials). Eve can then eavesdrop on the network, and because she knows her credentials, she can decrypt the SSV. Eve can compute a digest BIND_CONN_TO_SESSION will accept, and so bind a new connection to the session. Eve can change the slotid, sequence state, and/or the SSV state in such a way that when Bob accesses the server via the legitimate client, the legitimate client will be unable to use the session. The client's only recourse is to create a new session, which will cause any state Eve created on the legitimate client over the old (but hijacked) session to be lost. This disrupts Eve, but because she is the attacker, this is acceptable. Once the legitimate client establishes an SSV over the new session using Bob's RPCSEC_GSS context, Eve can use the new session via the legitimate client, but she cannot disrupt Bob. Moreover, because the client SHOULD have modified the SSV due to Eve using the new session, Bob cannot get revenge on Eve by binding a rogue connection to the session. The question is how does the legitimate client detect that Eve has hijacked the old session? When the client detects that a new principal, Bob, wants to use the session, it SHOULD have issued a SET_SSV. * Let us suppose that from the rogue connection, Eve issued a SET_SSV with the same slotid and sequence that the legitimate client later uses. The server will assume this is a replay, and return to the legitimate client the reply it sent Eve. However, unless Eve can correctly guess the SSV the legitimate client will use, the digest verification checks in the SET_SSV response will fail. That is the clue to the client that the session has been hijacked. Shepler, et al. Expires September 5, 2007 [Page 56] Internet-Draft NFSv4 Minor Version 1 March 2007 * Alternatively, Eve issued a SET_SSV with a different slotid than the legitimate client uses for its SET_SSV. Then the digest verification on the server fails, and the client is again clued that the session has been hijacked. * Alternatively, Eve issued an operation other than SET_SSV, but with the same slotid and sequence that the legitimate client uses for its SET_SSV. The server returns to the legitimate client the response it sent Eve. The client sees that the response is not at all what it expects. The client assumes either session hijacking or server bug, and either way destroys the old session. o Eve binds a rogue connection to the session as above, and then destroys the session. Again, Bob goes to use the server from the legitimate client. The client has a very clear indication that its session was hijacked, and does not even have to destroy the old session before creating a new session, which Eve will be unable to hijack because it will be protected with an SSV created via Bob's RPCSEC_GSS protection. o If Eve creates a connection before the legitimate client establishes an SSV, because the initial value of the SSV is zero and therefore known, Eve can issue a SET_SSV that will pass the digest verification check. However because the new connection has not been bound to the session, the SET_SSV is rejected for that reason. o The connection to session binding model does not prevent connection hijacking. However, if an attacker can perform connection hijacking, it can issue denial of service attacks that are less difficult than attacks based on forging sessions. 2.10.7. Session Mechanics - Steady State 2.10.7.1. Obligations of the Server The server has the primary obligation to monitor the state of backchannel resources that the client has created for the server (RPCSEC_GSS contexts and back channel connections). When these resources go away, the server takes action as specified in Section 2.10.8.2. 2.10.7.2. Obligations of the Client The client has the following obligations in order to utilize the session: Shepler, et al. Expires September 5, 2007 [Page 57] Internet-Draft NFSv4 Minor Version 1 March 2007 o Keep a necessary session from going idle on the server. A client that requires a session, but nonetheless is not sending operations risks having the session be destroyed by the server. This is because sessions consume resources, and resource limitations may force the server to cull the least recently used session. o Destroy the session when idle. When a session has no state other than the session, and no outstanding requests, the client should consider destroying the session. o Maintain GSS contexts for callback. If the client requires the server to use the RPCSEC_GSS security flavor for callbacks, then it needs to be sure the contexts handed to the server via BACKCHANNEL_CTL are unexpired. A good practice is to keep at least two contexts outstanding, where the expiration time of the newest context at the time it was created, is N times that of the oldest context, where N is the number of contexts available for callbacks. o Maintain an active connection. The server requires a callback path in order to gracefully recall recallable state, or notify the client of certain events. 2.10.7.3. Steps the Client Takes To Establish a Session The client issues EXCHANGE_ID to establish a client ID. The client uses the client ID to issue a CREATE_SESSION on a connection to the server. The results of CREATE_SESSION indicate whether the server will persist the session replay cache through a server reboot or not, and the client notes this for future reference. The client SHOULD have specified connecting binding enforcement when the session was created. If so, the client SHOULD issue SET_SSV in the first COMPOUND after the session is created. If it is not using machine credentials, then each time a new principal goes to use the session, it SHOULD issue a SET_SSV again. If the client wants to use delegations, layouts, directory notifications, or any other state that requires a callback channel, then it MUST add a connection to the backchannel if CREATE_SESSION did not already do so. The client creates a connection, and calls BIND_CONN_TO_SESSION to bind the connection to the session and the session's backchannel. If CREATE_SESSION did not already do so, the client MUST tell the server what security is required in order for the client to accept callbacks. The client does this via BACKCHANNEL_CTL. Shepler, et al. Expires September 5, 2007 [Page 58] Internet-Draft NFSv4 Minor Version 1 March 2007 If the client wants to use additional connections for the backchannel, then it MUST call BIND_CONN_TO_SESSION on each connection it wants to use with the session. If the client wants to use additional connections for the operation channel, then it MUST call BIND_CONN_TO_SESSION if it specified connection binding enforcement before using the connection. At this point the client has reached a steady state as far as session use. 2.10.8. Session Mechanics - Recovery 2.10.8.1. Events Requiring Client Action The following events require client action to recover. 2.10.8.1.1. RPCSEC_GSS Context Loss by Callback Path If all RPCSEC_GSS contexts granted to by the client to the server for callback use have expired, the client MUST establish a new context via BACKCHANNEL_CTL. The sr_status_flags field of the SEQUENCE results indicates when callback contexts are nearly expired, or fully expired (see Section 17.46.4). 2.10.8.1.2. Connection Disconnect If the client loses the last connection of the session, then it MUST create a new connection, and if connecting binding enforcement was specified when the session was created, bind it to the session via BIND_CONN_TO_SESSION. If there were requests outstanding at the time the of connection disconnect, then the client MUST retry the request, as described in Section 2.10.4.2. Note that it is not necessary to retry requests over a connection with the same source network address or the same destination network address as the disconnected connection. As long as the sessionid, slotid, and sequenceid in the retry match that of the original request, the server will recognize the request as a retry if it did see the request prior to disconnect. If the connection that was bound to the backchannel is lost, the client may need to reconnect, and use BIND_CONN_TO_SESSION, to give the connection to the backchannel. If the connection that was lost was the last one bound to the backchannel, the client MUST reconnect, and bind the connection to the session and backchannel. The server should indicate when it has no callback connection via the sr_status_flags result from SEQUENCE. Shepler, et al. Expires September 5, 2007 [Page 59] Internet-Draft NFSv4 Minor Version 1 March 2007 2.10.8.1.3. Backchannel GSS Context Loss Via the sr_status_flags result of the SEQUENCE operation or other means, the client will learn if some or all of the RPCSEC_GSS contexts it assigned to the backchannel have been lost. The client may need to use BACKCHANNEL_CTL to assign new contexts. It MUST assign new contexts if there are no more contexts. 2.10.8.1.4. Loss of Session The server may lose a record of the session. Causes include: o Server crash and reboot o A catastrophe that causes the cache to be corrupted or lost on the media it was stored on. This applies even if the server indicated in the CREATE_SESSION results that it would persist the cache. o The server purges the session of a client that has been inactive for a very extended period of time. [[Comment.11: XXX - Should we add a value to the CREATE_SESSION results that tells a client how long he can let a session stay idle before losing it?]] Loss of replay cache is equivalent to loss of session. The server indicates loss of session to the client by returning NFS4ERR_BADSESSION on the next operation that uses the sessionid associated with the lost session. After an event like a server reboot, the client may have lost its connections. The client assumes for the moment that the session has not been lost. It reconnects, and if it specified connecting binding enforcement when the session was created, it invokes BIND_CONN_TO_SESSION using the sessionid. Otherwise, it invokes SEQUENCE. If BIND_CONN_TO_SESSION or SEQUENCE returns NFS4ERR_BADSESSION, the client knows the session was lost. If the connection survives session loss, then the next SEQUENCE operation the client issues over the connection will get back NFS4ERR_BADSESSION. The client again knows the session was lost. When the client detects session loss, it must call CREATE_SESSION to recover. Any non-idempotent operations that were in progress may have been performed on the server at the time of session loss. The client has no general way to recover from this. Note that loss of session does not imply loss of lock, open, delegation, or layout state. Nor does loss of lock, open, delegation, or layout state imply loss of session state. [[Comment.12: Add reference to lock recovery section]] . A session Shepler, et al. Expires September 5, 2007 [Page 60] Internet-Draft NFSv4 Minor Version 1 March 2007 can survive a server reboot, but lock recovery may still be needed. The converse is also true. It is possible CREATE_SESSION will fail with NFS4ERR_STALE_CLIENTID (for example the server reboots and does not preserve client ID state). If so, the client needs to call EXCHANGE_ID, followed by CREATE_SESSION. 2.10.8.1.5. Failover [[Comment.13: Dave Noveck requested this section; not sure what is needed here if this refers to failover to a replica. What are the session ramifications?]] 2.10.8.2. Events Requiring Server Action The following events require server action to recover. 2.10.8.2.1. Client Crash and Reboot As described in Section 17.35, a rebooted client causes the server to delete any sessions it had. 2.10.8.2.2. Client Crash with No Reboot If a client crashes and never comes back, it will never issue EXCHANGE_ID with its old client owner. Thus the server has session state that will never be used again. After an extended period of time and if the server has resource constraints, it MAY destroy the old session. 2.10.8.2.3. Extended Network Partition To the server, the extended network partition may be no different than a client crash with no reboot (see Section 2.10.8.2.2). Unless the server can discern that there is a network partition, it is free to treat the situation as if the client has crashed for good. 2.10.8.2.4. Backchannel Connection Loss If there were callback requests outstanding at the time the of a connection disconnect, then the server MUST retry the request, as described in Section 2.10.4.2. Note that it is not necessary to retry requests over a connection with the same source network address or the same destination network address as the disconnected connection. As long as the sessionid, slotid, and sequenceid in the retry match that of the original request, the callback target will recognize the request as a retry if it did see the request prior to Shepler, et al. Expires September 5, 2007 [Page 61] Internet-Draft NFSv4 Minor Version 1 March 2007 disconnect. If the connection lost is the last one bound to the backchannel, then the server MUST indicate that in the sr_status_flags field of the next SEQUENCE reply. 2.10.8.2.5. GSS Context Loss The server SHOULD monitor when the last RPCSEC_GSS context assigned to the backchannel is near expiry (i.e. between one and two periods of lease time), and indicate so in the sr_status_flags field of the next SEQUENCE reply. The server MUST indicate when the backchannel's last RPCSEC_GSS context has expired in the sr_status_flags field of the next SEQUENCE reply. 2.10.9. Parallel NFS and Sessions A client and server can potentially be a non-pNFS implementation, a metadata server implementation, a data server implementation, or two or three types of implementations. The EXCHGID4_FLAG_USE_NON_PNFS, EXCHGID4_FLAG_USE_PNFS_MDS, and EXCHGID4_FLAG_USE_PNFS_DS flags (not mutually exclusive) are passed in the EXCHANGE_ID arguments and results to allow the client to indicate how it wants to use sessions created under the client ID, and to allow the server to indicate how it will allow the sessions to be used. See Section 13.1 for pNFS sessions considerations. 3. Protocol Data Types The syntax and semantics to describe the data types of the NFS version 4 protocol are defined in the XDR RFC4506 [3] and RPC RFC1831 [4] documents. The next sections build upon the XDR data types to define types and structures specific to this protocol. 3.1. Basic Data Types These are the base NFSv4 data types. +---------------+---------------------------------------------------+ | Data Type | Definition | +---------------+---------------------------------------------------+ | int32_t | typedef int int32_t; | | uint32_t | typedef unsigned int uint32_t; | | int64_t | typedef hyper int64_t; | | uint64_t | typedef unsigned hyper uint64_t; | | attrlist4 | typedef opaque attrlist4<>; | | | Used for file/directory attributes | Shepler, et al. Expires September 5, 2007 [Page 62] Internet-Draft NFSv4 Minor Version 1 March 2007 | bitmap4 | typedef uint32_t bitmap4<>; | | | Used in attribute array encoding. | | changeid4 | typedef uint64_t changeid4; | | | Used in definition of change_info | | clientid4 | typedef uint64_t clientid4; | | | Shorthand reference to client identification | | component4 | typedef utf8str_cs component4; | | | Represents path name components | | count4 | typedef uint32_t count4; | | | Various count parameters (READ, WRITE, COMMIT) | | length4 | typedef uint64_t length4; | | | Describes LOCK lengths | | linktext4 | typedef utf8str_cs linktext4; | | | Symbolic link contents | | mode4 | typedef uint32_t mode4; | | | Mode attribute data type | | nfs_cookie4 | typedef uint64_t nfs_cookie4; | | | Opaque cookie value for READDIR | | nfs_fh4 | typedef opaque nfs_fh4 | | | Filehandle definition; NFS4_FHSIZE is defined as | | | 128 | | nfs_ftype4 | enum nfs_ftype4; | | | Various defined file types | | nfsstat4 | enum nfsstat4; | | | Return value for operations | | offset4 | typedef uint64_t offset4; | | | Various offset designations (READ, WRITE, LOCK, | | | COMMIT) | | pathname4 | typedef component4 pathname4<>; | | | Represents path name for fs_locations | | qop4 | typedef uint32_t qop4; | | | Quality of protection designation in SECINFO | | sec_oid4 | typedef opaque sec_oid4<>; | | | Security Object Identifier The sec_oid4 data type | | | is not really opaque. Instead contains an ASN.1 | | | OBJECT IDENTIFIER as used by GSS-API in the | | | mech_type argument to GSS_Init_sec_context. See | | | RFC2743 [8] for details. | | sequenceid4 | typedef uint32_t sequenceid4; | | | sequence number used for various session | | | operations (EXCHANGE_ID, CREATE_SESSION, | | | SEQUENCE, CB_SEQUENCE). | | seqid4 | typedef uint32_t seqid4; | | | Sequence identifier used for file locking | | sessionid4 | typedef opaque sessionid4[16]; | | | Session identifier | | slotid4 | typedef uint32_t slotid4; | Shepler, et al. Expires September 5, 2007 [Page 63] Internet-Draft NFSv4 Minor Version 1 March 2007 | | sequencing artifact various session operations | | | (SEQUENCE, CB_SEQUENCE). | | utf8string | typedef opaque utf8string<>; | | | UTF-8 encoding for strings | | utf8str_cis | typedef opaque utf8str_cis; | | | Case-insensitive UTF-8 string | | utf8str_cs | typedef opaque utf8str_cs; | | | Case-sensitive UTF-8 string | | utf8str_mixed | typedef opaque utf8str_mixed; | | | UTF-8 strings with a case sensitive prefix and a | | | case insensitive suffix. | | verifier4 | typedef opaque verifier4[NFS4_VERIFIER_SIZE]; | | | Verifier used for various operations (COMMIT, | | | CREATE, EXCHANGE_ID, OPEN, READDIR, WRITE) | | | NFS4_VERIFIER_SIZE is defined as 8. | +---------------+---------------------------------------------------+ End of Base Data Types Table 1 3.2. Structured Data Types 3.2.1. nfstime4 struct nfstime4 { int64_t seconds; uint32_t nseconds; } The nfstime4 structure gives the number of seconds and nanoseconds since midnight or 0 hour January 1, 1970 Coordinated Universal Time (UTC). Values greater than zero for the seconds field denote dates after the 0 hour January 1, 1970. Values less than zero for the seconds field denote dates before the 0 hour January 1, 1970. In both cases, the nseconds field is to be added to the seconds field for the final time representation. For example, if the time to be represented is one-half second before 0 hour January 1, 1970, the seconds field would have a value of negative one (-1) and the nseconds fields would have a value of one-half second (500000000). Values greater than 999,999,999 for nseconds are considered invalid. This data type is used to pass time and date information. A server converts to and from its local representation of time when processing time values, preserving as much accuracy as possible. If the precision of timestamps stored for a file system object is less than defined, loss of precision can occur. An adjunct time maintenance protocol is recommended to reduce client and server time skew. Shepler, et al. Expires September 5, 2007 [Page 64] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.2. time_how4 enum time_how4 { SET_TO_SERVER_TIME4 = 0, SET_TO_CLIENT_TIME4 = 1 }; 3.2.3. settime4 union settime4 switch (time_how4 set_it) { case SET_TO_CLIENT_TIME4: nfstime4 time; default: void; }; The above definitions are used as the attribute definitions to set time values. If set_it is SET_TO_SERVER_TIME4, then the server uses its local representation of time for the time value. 3.2.4. specdata4 struct specdata4 { uint32_t specdata1; /* major device number */ uint32_t specdata2; /* minor device number */ }; This data type represents additional information for the device file types NF4CHR and NF4BLK. 3.2.5. fsid4 struct fsid4 { uint64_t major; uint64_t minor; }; 3.2.6. fs_location4 struct fs_location4 { utf8str_cis server<>; pathname4 rootpath; }; Shepler, et al. Expires September 5, 2007 [Page 65] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.7. fs_locations4 struct fs_locations4 { pathname4 fs_root; fs_location4 locations<>; }; The fs_location4 and fs_locations4 data types are used for the fs_locations recommended attribute which is used for migration and replication support. 3.2.8. fattr4 struct fattr4 { bitmap4 attrmask; attrlist4 attr_vals; }; The fattr4 structure is used to represent file and directory attributes. The bitmap is a counted array of 32 bit integers used to contain bit values. The position of the integer in the array that contains bit n can be computed from the expression (n / 32) and its bit within that integer is (n mod 32). 0 1 +-----------+-----------+-----------+-- | count | 31 .. 0 | 63 .. 32 | +-----------+-----------+-----------+-- 3.2.9. change_info4 struct change_info4 { bool atomic; changeid4 before; changeid4 after; }; This structure is used with the CREATE, LINK, REMOVE, RENAME operations to let the client know the value of the change attribute for the directory in which the target file system object resides. Shepler, et al. Expires September 5, 2007 [Page 66] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.10. netaddr4 struct netaddr4 { /* see struct rpcb in RFC1833 */ string r_netid<>; /* network id */ string r_addr<>; /* universal address */ }; The netaddr4 structure is used to identify TCP/IP based endpoints. The r_netid and r_addr fields are specified in RFC1833 [22], but they are underspecified in RFC1833 [22] as far as what they should look like for specific protocols. For TCP over IPv4 and for UDP over IPv4, the format of r_addr is the US-ASCII string: h1.h2.h3.h4.p1.p2 The prefix, "h1.h2.h3.h4", is the standard textual form for representing an IPv4 address, which is always four octets long. Assuming big-endian ordering, h1, h2, h3, and h4, are respectively, the first through fourth octets each converted to ASCII-decimal. Assuming big-endian ordering, p1 and p2 are, respectively, the first and second octets each converted to ASCII-decimal. For example, if a host, in big-endian order, has an address of 0x0A010307 and there is a service listening on, in big endian order, port 0x020F (decimal 527), then complete universal address is "10.1.3.7.2.15". For TCP over IPv4 the value of r_netid is the string "tcp". For UDP over IPv4 the value of r_netid is the string "udp". That this document specifies the universal address and netid for UDP/IPv6 does not imply that UDP/IPv4 is a legal transport for NFSv4.1 (see Section 2.9). For TCP over IPv6 and for UDP over IPv6, the format of r_addr is the US-ASCII string: x1:x2:x3:x4:x5:x6:x7:x8.p1.p2 The suffix "p1.p2" is the service port, and is computed the same way as with universal addresses for TCP and UDP over IPv4. The prefix, "x1:x2:x3:x4:x5:x6:x7:x8", is the standard textual form for representing an IPv6 address as defined in Section 2.2 of RFC1884 [9]. Additionally, the two alternative forms specified in Section 2.2 of RFC1884 [9] are also acceptable. For TCP over IPv6 the value of r_netid is the string "tcp6". For UDP over IPv6 the value of r_netid is the string "udp6". That this Shepler, et al. Expires September 5, 2007 [Page 67] Internet-Draft NFSv4 Minor Version 1 March 2007 document specifies the universal address and netid for UDP/IPv6 does not imply that UDP/IPv6 is a legal transport for NFSv4.1 (see Section 2.9). 3.2.11. open_owner4 struct open_owner4 { clientid4 clientid; opaque owner }; This structure is used to identify the owner of open state. NFS4_OPAQUE_LIMIT is defined as 1024. 3.2.12. lock_owner4 struct lock_owner4 { clientid4 clientid; opaque owner }; This structure is used to identify the owner of file locking state. 3.2.13. open_to_lock_owner4 struct open_to_lock_owner4 { seqid4 open_seqid; stateid4 open_stateid; seqid4 lock_seqid; lock_owner4 lock_owner; }; This structure is used for the first LOCK operation done for an open_owner4. It provides both the open_stateid and lock_owner such that the transition is made from a valid open_stateid sequence to that of the new lock_stateid sequence. Using this mechanism avoids the confirmation of the lock_owner/lock_seqid pair since it is tied to established state in the form of the open_stateid/open_seqid. 3.2.14. stateid4 struct stateid4 { uint32_t seqid; opaque other[12]; }; This structure is used for the various state sharing mechanisms between the client and server. For the client, this data structure Shepler, et al. Expires September 5, 2007 [Page 68] Internet-Draft NFSv4 Minor Version 1 March 2007 is read-only. The starting value of the seqid field is undefined. The server is required to increment the seqid field monotonically at each transition of the stateid. This is important since the client will inspect the seqid in OPEN stateids to determine the order of OPEN processing done by the server. 3.2.15. layouttype4 enum layouttype4 { LAYOUT4_NFSV4_1_FILES = 1, LAYOUT4_OSD2_OBJECTS = 2, LAYOUT4_BLOCK_VOLUME = 3 }; A layout type specifies the layout being used. The implication is that clients have "layout drivers" that support one or more layout types. The file server advertises the layout types it supports through the fs_layout_type file system attribute (Section 5.13.1). A client asks for layouts of a particular type in LAYOUTGET, and passes those layouts to its layout driver. The layouttype4 structure is 32 bits in length. The range represented by the layout type is split into three parts. Type 0x0 is reserved. Types within the range 0x00000001-0x7FFFFFFF are globally unique and are assigned according to the description in Section 21.1; they are maintained by IANA. Types within the range 0x80000000-0xFFFFFFFF are site specific and for "private use" only. The LAYOUT4_NFSV4_1_FILES enumeration specifies that the NFSv4.1 file layout type is to be used. The LAYOUT4_OSD2_OBJECTS enumeration specifies that the object layout, as defined in [23], is to be used. Similarly, the LAYOUT4_BLOCK_VOLUME enumeration that the block/volume layout, as defined in [24], is to be used. 3.2.16. deviceid4 typedef uint32_t deviceid4; /* 32-bit device ID */ Layout information includes device IDs that specify a storage device through a compact handle. Addressing and type information is obtained with the GETDEVICEINFO operation. A client must not assume that device IDs are valid across metadata server reboots. The device ID is qualified by the layout type and are unique per file system (FSID). This allows different layout drivers to generate device IDs without the need for co-ordination. See Section 12.2.12 for more details. Shepler, et al. Expires September 5, 2007 [Page 69] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.17. device_addr4 struct device_addr4 { layouttype4 da_layout_type; opaque da_addr_body<>; }; The device address is used to set up a communication channel with the storage device. Different layout types will require different types of structures to define how they communicate with storage devices. The opaque da_addr_body field must be interpreted based on the specified da_layout_type field. This document defines the device address for the NFSv4.1 file layout ([[Comment.14: need xref]]), which identifies a storage device by network IP address and port number. This is sufficient for the clients to communicate with the NFSv4.1 storage devices, and may be sufficient for other layout types as well. Device types for object storage devices and block storage devices (e.g., SCSI volume labels) will be defined by their respective layout specifications. 3.2.18. devlist_item4 struct devlist_item4 { deviceid4 dli_id; device_addr4 dli_device_addr<>; }; An array of these values is returned by the GETDEVICELIST operation. They define the set of devices associated with a file system for the layout type specified in the GETDEVICELIST4args. 3.2.19. layout_content4 struct layout_content4 { layouttype4 loc_type; opaque loc_body<>; }; The loc_body field must be interpreted based on the layout type (loc_type). This document defines the loc_body for the NFSv4.1 file layout type is defined; see Section 13.3 for its definition. Shepler, et al. Expires September 5, 2007 [Page 70] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.20. layout4 struct layout4 { offset4 lo_offset; length4 lo_length; layoutiomode4 lo_iomode; layout_content4 lo_content; }; The layout4 structure defines a layout for a file. The layout type specific data is opaque within lo_content. Since layouts are sub- dividable, the offset and length together with the file's filehandle, the client ID, iomode, and layout type, identifies the layout. 3.2.21. layoutupdate4 struct layoutupdate4 { layouttype4 lou_type; opaque lou_body<>; }; The layoutupdate4 structure is used by the client to return 'updated' layout information to the metadata server at LAYOUTCOMMIT time. This structure provides a channel to pass layout type specific information (in field lou_body) back to the metadata server. E.g., for block/ volume layout types this could include the list of reserved blocks that were written. The contents of the opaque lou_body argument are determined by the layout type and are defined in their context. The NFSv4.1 file-based layout does not use this structure, thus the lou_body field should have a zero length. 3.2.22. layouthint4 struct layouthint4 { layouttype4 loh_type; opaque loh_body<>; }; The layouthint4 structure is used by the client to pass in a hint about the type of layout it would like created for a particular file. It is the structure specified by the layout_hint attribute described in Section 5.13.4. The metadata server may ignore the hint, or may selectively ignore fields within the hint. This hint should be provided at create time as part of the initial attributes within OPEN. The loh_body field is specific to the type of layout (loh_type). The NFSv4.1 file-based layout uses the nfsv4_1_file_layouthint4 structure as defined in Section 13.3. Shepler, et al. Expires September 5, 2007 [Page 71] Internet-Draft NFSv4 Minor Version 1 March 2007 3.2.23. layoutiomode4 enum layoutiomode4 { LAYOUTIOMODE4_READ = 1, LAYOUTIOMODE4_RW = 2, LAYOUTIOMODE4_ANY = 3 }; The iomode specifies whether the client intends to read or write (with the possibility of reading) the data represented by the layout. The ANY iomode MUST NOT be used for LAYOUTGET, however, it can be used for LAYOUTRETURN and LAYOUTRECALL. The ANY iomode specifies that layouts pertaining to both READ and RW iomodes are being returned or recalled, respectively. The metadata server's use of the iomode may depend on the layout type being used. The storage devices may validate I/O accesses against the iomode and reject invalid accesses. 3.2.24. nfs_impl_id4 struct nfs_impl_id4 { utf8str_cis nii_domain; utf8str_cs nii_name; nfstime4 nii_date; }; This structure is used to identify client and server implementation detail. The nii_domain field is the DNS domain name that the implementer is associated with. The nii_name field is the product name of the implementation and is completely free form. It is recommended that the nii_name be used to distinguish machine architecture, machine platforms, revisions, versions, and patch levels. The nii_date field is the timestamp of when the software instance was published or built. 3.2.25. threshold_item4 struct threshold_item4 { layouttype4 thi_layout_type; bitmap4 thi_hintset; opaque thi_hintlist<>; }; This structure contains a list of hints specific to a layout type for helping the client determine when it should issue I/O directly through the metadata server vs. the data servers. The hint structure consists of the layout type (thi_layout_type), a bitmap (thi_hintset) describing the set of hints supported by the server (they may differ Shepler, et al. Expires September 5, 2007 [Page 72] Internet-Draft NFSv4 Minor Version 1 March 2007 based on the layout type), and a list of hints (thi_hintlist), whose structure is determined by the hintset bitmap. See the mdsthreshold attribute for more details. The thi_hintset field is a bitmap of the following values: +-------------------------+---+---------+---------------------------+ | name | # | Data | Description | | | | Type | | +-------------------------+---+---------+---------------------------+ | threshold4_read_size | 0 | length4 | The file size below which | | | | | it is recommended to read | | | | | data through the MDS. | | threshold4_write_size | 1 | length4 | The file size below which | | | | | it is recommended to | | | | | write data through the | | | | | MDS. | | threshold4_read_iosize | 2 | length4 | For read I/O sizes below | | | | | this threshold it is | | | | | recommended to read data | | | | | through the MDS | | threshold4_write_iosize | 3 | length4 | For write I/O sizes below | | | | | this threshold it is | | | | | recommended to write data | | | | | through the MDS | +-------------------------+---+---------+---------------------------+ 3.2.26. mdsthreshold4 struct mdsthreshold4 { threshold_item4 mth_hints<>; }; This structure holds an array of threshold_item4 structures each of which is valid for a particular layout type. An array is necessary since a server can support multiple layout types for a single file. 4. Filehandles The filehandle in the NFS protocol is a per server unique identifier for a file system object. The contents of the filehandle are opaque to the client. Therefore, the server is responsible for translating the filehandle to an internal representation of the file system object. Shepler, et al. Expires September 5, 2007 [Page 73] Internet-Draft NFSv4 Minor Version 1 March 2007 4.1. Obtaining the First Filehandle The operations of the NFS protocol are defined in terms of one or more filehandles. Therefore, the client needs a filehandle to initiate communication with the server. With the NFS version 2 protocol RFC1094 [17] and the NFS version 3 protocol RFC1813 [18], there exists an ancillary protocol to obtain this first filehandle. The MOUNT protocol, RPC program number 100005, provides the mechanism of translating a string based file system path name to a filehandle which can then be used by the NFS protocols. The MOUNT protocol has deficiencies in the area of security and use via firewalls. This is one reason that the use of the public filehandle was introduced in RFC2054 [25] and RFC2055 [26]. With the use of the public filehandle in combination with the LOOKUP operation in the NFS version 2 and 3 protocols, it has been demonstrated that the MOUNT protocol is unnecessary for viable interaction between NFS client and server. Therefore, the NFS version 4 protocol will not use an ancillary protocol for translation from string based path names to a filehandle. Two special filehandles will be used as starting points for the NFS client. 4.1.1. Root Filehandle The first of the special filehandles is the ROOT filehandle. The ROOT filehandle is the "conceptual" root of the file system name space at the NFS server. The client uses or starts with the ROOT filehandle by employing the PUTROOTFH operation. The PUTROOTFH operation instructs the server to set the "current" filehandle to the ROOT of the server's file tree. Once this PUTROOTFH operation is used, the client can then traverse the entirety of the server's file tree with the LOOKUP operation. A complete discussion of the server name space is in the section "NFS Server Name Space". 4.1.2. Public Filehandle The second special filehandle is the PUBLIC filehandle. Unlike the ROOT filehandle, the PUBLIC filehandle may be bound or represent an arbitrary file system object at the server. The server is responsible for this binding. It may be that the PUBLIC filehandle and the ROOT filehandle refer to the same file system object. However, it is up to the administrative software at the server and the policies of the server administrator to define the binding of the PUBLIC filehandle and server file system object. The client may not make any assumptions about this binding. The client uses the PUBLIC filehandle via the PUTPUBFH operation. Shepler, et al. Expires September 5, 2007 [Page 74] Internet-Draft NFSv4 Minor Version 1 March 2007 4.2. Filehandle Types In the NFS version 2 and 3 protocols, there was one type of filehandle with a single set of semantics. This type of filehandle is termed "persistent" in NFS Version 4. The semantics of a persistent filehandle remain the same as before. A new type of filehandle introduced in NFS Version 4 is the "volatile" filehandle, which attempts to accommodate certain server environments. The volatile filehandle type was introduced to address server functionality or implementation issues which make correct implementation of a persistent filehandle infeasible. Some server environments do not provide a file system level invariant that can be used to construct a persistent filehandle. The underlying server file system may not provide the invariant or the server's file system programming interfaces may not provide access to the needed invariant. Volatile filehandles may ease the implementation of server functionality such as hierarchical storage management or file system reorganization or migration. However, the volatile filehandle increases the implementation burden for the client. Since the client will need to handle persistent and volatile filehandles differently, a file attribute is defined which may be used by the client to determine the filehandle types being returned by the server. 4.2.1. General Properties of a Filehandle The filehandle contains all the information the server needs to distinguish an individual file. To the client, the filehandle is opaque. The client stores filehandles for use in a later request and can compare two filehandles from the same server for equality by doing an octet-by-octet comparison. However, the client MUST NOT otherwise interpret the contents of filehandles. If two filehandles from the same server are equal, they MUST refer to the same file. Servers SHOULD try to maintain a one-to-one correspondence between filehandles and files but this is not required. Clients MUST use filehandle comparisons only to improve performance, not for correct behavior. All clients need to be prepared for situations in which it cannot be determined whether two filehandles denote the same object and in such cases, avoid making invalid assumptions which might cause incorrect behavior. Further discussion of filehandle and attribute comparison in the context of data caching is presented in the section "Data Caching and File Identity". As an example, in the case that two different path names when traversed at the server terminate at the same file system object, the server SHOULD return the same filehandle for each path. This can Shepler, et al. Expires September 5, 2007 [Page 75] Internet-Draft NFSv4 Minor Version 1 March 2007 occur if a hard link is used to create two file names which refer to the same underlying file object and associated data. For example, if paths /a/b/c and /a/d/c refer to the same file, the server SHOULD return the same filehandle for both path names traversals. 4.2.2. Persistent Filehandle A persistent filehandle is defined as having a fixed value for the lifetime of the file system object to which it refers. Once the server creates the filehandle for a file system object, the server MUST accept the same filehandle for the object for the lifetime of the object. If the server restarts or reboots the NFS server must honor the same filehandle value as it did in the server's previous instantiation. Similarly, if the file system is migrated, the new NFS server must honor the same filehandle as the old NFS server. The persistent filehandle will be become stale or invalid when the file system object is removed. When the server is presented with a persistent filehandle that refers to a deleted object, it MUST return an error of NFS4ERR_STALE. A filehandle may become stale when the file system containing the object is no longer available. The file system may become unavailable if it exists on removable media and the media is no longer available at the server or the file system in whole has been destroyed or the file system has simply been removed from the server's name space (i.e. unmounted in a UNIX environment). 4.2.3. Volatile Filehandle A volatile filehandle does not share the same longevity characteristics of a persistent filehandle. The server may determine that a volatile filehandle is no longer valid at many different points in time. If the server can definitively determine that a volatile filehandle refers to an object that has been removed, the server should return NFS4ERR_STALE to the client (as is the case for persistent filehandles). In all other cases where the server determines that a volatile filehandle can no longer be used, it should return an error of NFS4ERR_FHEXPIRED. The mandatory attribute "fh_expire_type" is used by the client to determine what type of filehandle the server is providing for a particular file system. This attribute is a bitmask with the following values: FH4_PERSISTENT The value of FH4_PERSISTENT is used to indicate a persistent filehandle, which is valid until the object is removed from the file system. The server will not return NFS4ERR_FHEXPIRED for this filehandle. FH4_PERSISTENT is defined as a value in which none of the bits specified below are set. Shepler, et al. Expires September 5, 2007 [Page 76] Internet-Draft NFSv4 Minor Version 1 March 2007 FH4_VOLATILE_ANY The filehandle may expire at any time, except as specifically excluded (i.e. FH4_NO_EXPIRE_WITH_OPEN). FH4_NOEXPIRE_WITH_OPEN May only be set when FH4_VOLATILE_ANY is set. If this bit is set, then the meaning of FH4_VOLATILE_ANY is qualified to exclude any expiration of the filehandle when it is open. FH4_VOL_MIGRATION The filehandle will expire as a result of a file system transition (migration or replication), in those case in which the continuity of filehandle use is not specified by _handle_ class information within the fs_locations_info attribute. When this bit is set, clients without access to fs_locations_info information should assume filehandles will expire on file system transitions. FH4_VOL_RENAME The filehandle will expire during rename. This includes a rename by the requesting client or a rename by any other client. If FH4_VOL_ANY is set, FH4_VOL_RENAME is redundant. Servers which provide volatile filehandles that may expire while open (i.e. if FH4_VOL_MIGRATION or FH4_VOL_RENAME is set or if FH4_VOLATILE_ANY is set and FH4_NOEXPIRE_WITH_OPEN not set), should deny a RENAME or REMOVE that would affect an OPEN file of any of the components leading to the OPEN file. In addition, the server should deny all RENAME or REMOVE requests during the grace period upon server restart. Servers which provide volatile filehandles that may expire while open require special care as regards handling of RENAMESs and REMOVEs. This situation can arise if FH4_VOL_MIGRATION or FH4_VOL_RENAME is set, if FH4_VOLATILE_ANY is set and FH4_NOEXPIRE_WITH_OPEN not set, or if a non-readonly file system has a transition target in a different _handle _ class. In these cases, the server should deny a RENAME or REMOVE that would affect an OPEN file of any of the components leading to the OPEN file. In addition, the server should deny all RENAME or REMOVE requests during the grace period, in order to make sure that reclaims of files where filehandles may have expired do not do a reclaim for the wrong file. 4.3. One Method of Constructing a Volatile Filehandle A volatile filehandle, while opaque to the client could contain: [volatile bit = 1 | server boot time | slot | generation number] Shepler, et al. Expires September 5, 2007 [Page 77] Internet-Draft NFSv4 Minor Version 1 March 2007 o slot is an index in the server volatile filehandle table o generation number is the generation number for the table entry/ slot When the client presents a volatile filehandle, the server makes the following checks, which assume that the check for the volatile bit has passed. If the server boot time is less than the current server boot time, return NFS4ERR_FHEXPIRED. If slot is out of range, return NFS4ERR_BADHANDLE. If the generation number does not match, return NFS4ERR_FHEXPIRED. When the server reboots, the table is gone (it is volatile). If volatile bit is 0, then it is a persistent filehandle with a different structure following it. 4.4. Client Recovery from Filehandle Expiration If possible, the client SHOULD recover from the receipt of an NFS4ERR_FHEXPIRED error. The client must take on additional responsibility so that it may prepare itself to recover from the expiration of a volatile filehandle. If the server returns persistent filehandles, the client does not need these additional steps. For volatile filehandles, most commonly the client will need to store the component names leading up to and including the file system object in question. With these names, the client should be able to recover by finding a filehandle in the name space that is still available or by starting at the root of the server's file system name space. If the expired filehandle refers to an object that has been removed from the file system, obviously the client will not be able to recover from the expired filehandle. It is also possible that the expired filehandle refers to a file that has been renamed. If the file was renamed by another client, again it is possible that the original client will not be able to recover. However, in the case that the client itself is renaming the file and the file is open, it is possible that the client may be able to recover. The client can determine the new path name based on the processing of the rename request. The client can then regenerate the new filehandle based on the new path name. The client could also use the compound operation mechanism to construct a set of operations like: Shepler, et al. Expires September 5, 2007 [Page 78] Internet-Draft NFSv4 Minor Version 1 March 2007 RENAME A B LOOKUP B GETFH Note that the COMPOUND procedure does not provide atomicity. This example only reduces the overhead of recovering from an expired filehandle. 5. File Attributes To meet the requirements of extensibility and increased interoperability with non-UNIX platforms, attributes must be handled in a flexible manner. The NFS version 3 fattr3 structure contains a fixed list of attributes that not all clients and servers are able to support or care about. The fattr3 structure can not be extended as new needs arise and it provides no way to indicate non-support. With the NFS version 4 protocol, the client is able query what attributes the server supports and construct requests with only those supported attributes (or a subset thereof). To this end, attributes are divided into three groups: mandatory, recommended, and named. Both mandatory and recommended attributes are supported in the NFS version 4 protocol by a specific and well- defined encoding and are identified by number. They are requested by setting a bit in the bit vector sent in the GETATTR request; the server response includes a bit vector to list what attributes were returned in the response. New mandatory or recommended attributes may be added to the NFS protocol between major revisions by publishing a standards-track RFC which allocates a new attribute number value and defines the encoding for the attribute. See the section "Minor Versioning" for further discussion. Named attributes are accessed by the new OPENATTR operation, which accesses a hidden directory of attributes associated with a file system object. OPENATTR takes a filehandle for the object and returns the filehandle for the attribute hierarchy. The filehandle for the named attributes is a directory object accessible by LOOKUP or READDIR and contains files whose names represent the named attributes and whose data bytes are the value of the attribute. For example: Shepler, et al. Expires September 5, 2007 [Page 79] Internet-Draft NFSv4 Minor Version 1 March 2007 +----------+-----------+---------------------------------+ | LOOKUP | "foo" | ; look up file | | GETATTR | attrbits | | | OPENATTR | | ; access foo's named attributes | | LOOKUP | "x11icon" | ; look up specific attribute | | READ | 0,4096 | ; read stream of bytes | +----------+-----------+---------------------------------+ Named attributes are intended for data needed by applications rather than by an NFS client implementation. NFS implementors are strongly encouraged to define their new attributes as recommended attributes by bringing them to the IETF standards-track process. The set of attributes which are classified as mandatory is deliberately small since servers must do whatever it takes to support them. A server should support as many of the recommended attributes as possible but by their definition, the server is not required to support all of them. Attributes are deemed mandatory if the data is both needed by a large number of clients and is not otherwise reasonably computable by the client when support is not provided on the server. Note that the hidden directory returned by OPENATTR is a convenience for protocol processing. The client should not make any assumptions about the server's implementation of named attributes and whether the underlying file system at the server has a named attribute directory or not. Therefore, operations such as SETATTR and GETATTR on the named attribute directory are undefined. 5.1. Mandatory Attributes These MUST be supported by every NFS version 4 client and server in order to ensure a minimum level of interoperability. The server must store and return these attributes and the client must be able to function with an attribute set limited to these attributes. With just the mandatory attributes some client functionality may be impaired or limited in some ways. A client may ask for any of these attributes to be returned by setting a bit in the GETATTR request and the server must return their value. 5.2. Recommended Attributes These attributes are understood well enough to warrant support in the NFS version 4 protocol. However, they may not be supported on all clients and servers. A client may ask for any of these attributes to be returned by setting a bit in the GETATTR request but must handle the case where the server does not return them. A client may ask for the set of attributes the server supports and should not request Shepler, et al. Expires September 5, 2007 [Page 80] Internet-Draft NFSv4 Minor Version 1 March 2007 attributes the server does not support. A server should be tolerant of requests for unsupported attributes and simply not return them rather than considering the request an error. It is expected that servers will support all attributes they comfortably can and only fail to support attributes which are difficult to support in their operating environments. A server should provide attributes whenever they don't have to "tell lies" to the client. For example, a file modification time should be either an accurate time or should not be supported by the server. This will not always be comfortable to clients but the client is better positioned decide whether and how to fabricate or construct an attribute or whether to do without the attribute. 5.3. Named Attributes These attributes are not supported by direct encoding in the NFS Version 4 protocol but are accessed by string names rather than numbers and correspond to an uninterpreted stream of bytes which are stored with the file system object. The name space for these attributes may be accessed by using the OPENATTR operation. The OPENATTR operation returns a filehandle for a virtual "attribute directory" and further perusal of the name space may be done using READDIR and LOOKUP operations on this filehandle. Named attributes may then be examined or changed by normal READ and WRITE and CREATE operations on the filehandles returned from READDIR and LOOKUP. Named attributes may have attributes. It is recommended that servers support arbitrary named attributes. A client should not depend on the ability to store any named attributes in the server's file system. If a server does support named attributes, a client which is also able to handle them should be able to copy a file's data and meta-data with complete transparency from one location to another; this would imply that names allowed for regular directory entries are valid for named attribute names as well. Names of attributes will not be controlled by this document or other IETF standards track documents. See the section "IANA Considerations" for further discussion. 5.4. Classification of Attributes Each of the Mandatory and Recommended attributes can be classified in one of three categories: per server, per file system, or per file system object. Note that it is possible that some per file system attributes may vary within the file system. See the "homogeneous" attribute for its definition. Note that the attributes time_access_set and time_modify_set are not listed in this section Shepler, et al. Expires September 5, 2007 [Page 81] Internet-Draft NFSv4 Minor Version 1 March 2007 because they are write-only attributes corresponding to time_access and time_modify, and are used in a special instance of SETATTR. o The per server attribute is: lease_time o The per file system attributes are: supp_attr, fh_expire_type, link_support, symlink_support, unique_handles, aclsupport, cansettime, case_insensitive, case_preserving, chown_restricted, files_avail, files_free, files_total, fs_locations, homogeneous, maxfilesize, maxname, maxread, maxwrite, no_trunc, space_avail, space_free, space_total, time_delta, fs_status, fs_layout_type, fs_locations_info o The per file system object attributes are: type, change, size, named_attr, fsid, rdattr_error, filehandle, ACL, archive, fileid, hidden, maxlink, mimetype, mode, numlinks, owner, owner_group, rawdev, space_used, system, time_access, time_backup, time_create, time_metadata, time_modify, mounted_on_fileid, dir_notif_delay, dirent_notif_delay, dacl, sacl, layout_type, layout_hint, layout_blksize, layout_alignment, mdsthreshold, retention_get, retention_set, retentevt_get, retentevt_set, retention_hold, mode_set_masked For quota_avail_hard, quota_avail_soft, and quota_used see their definitions below for the appropriate classification. Shepler, et al. Expires September 5, 2007 [Page 82] Internet-Draft NFSv4 Minor Version 1 March 2007 5.5. Mandatory Attributes - Definitions +-----------------+----+------------+--------+----------------------+ | name | # | Data Type | Access | Description | +-----------------+----+------------+--------+----------------------+ | supp_attr | 0 | bitmap | READ | The bit vector which | | | | | | would retrieve all | | | | | | mandatory and | | | | | | recommended | | | | | | attributes that are | | | | | | supported for this | | | | | | object. The scope | | | | | | of this attribute | | | | | | applies to all | | | | | | objects with a | | | | | | matching fsid. | | type | 1 | nfs4_ftype | READ | The type of the | | | | | | object (file, | | | | | | directory, symlink, | | | | | | etc.) | | fh_expire_type | 2 | uint32 | READ | Server uses this to | | | | | | specify filehandle | | | | | | expiration behavior | | | | | | to the client. See | | | | | | the section | | | | | | "Filehandles" for | | | | | | additional | | | | | | description. | | change | 3 | uint64 | READ | A value created by | | | | | | the server that the | | | | | | client can use to | | | | | | determine if file | | | | | | data, directory | | | | | | contents or | | | | | | attributes of the | | | | | | object have been | | | | | | modified. The | | | | | | server may return | | | | | | the object's | | | | | | time_metadata | | | | | | attribute for this | | | | | | attribute's value | | | | | | but only if the file | | | | | | system object can | | | | | | not be updated more | | | | | | frequently than the | | | | | | resolution of | | | | | | time_metadata. | Shepler, et al. Expires September 5, 2007 [Page 83] Internet-Draft NFSv4 Minor Version 1 March 2007 | size | 4 | uint64 | R/W | The size of the | | | | | | object in bytes. | | link_support | 5 | bool | READ | True, if the | | | | | | object's file system | | | | | | supports hard links. | | symlink_support | 6 | bool | READ | True, if the | | | | | | object's file system | | | | | | supports symbolic | | | | | | links. | | named_attr | 7 | bool | READ | True, if this object | | | | | | has named | | | | | | attributes. In | | | | | | other words, object | | | | | | has a non-empty | | | | | | named attribute | | | | | | directory. | | fsid | 8 | fsid4 | READ | Unique file system | | | | | | identifier for the | | | | | | file system holding | | | | | | this object. fsid | | | | | | contains major and | | | | | | minor components | | | | | | each of which are | | | | | | uint64. | | unique_handles | 9 | bool | READ | True, if two | | | | | | distinct filehandles | | | | | | guaranteed to refer | | | | | | to two different | | | | | | file system objects. | | lease_time | 10 | nfs_lease4 | READ | Duration of leases | | | | | | at server in | | | | | | seconds. | | rdattr_error | 11 | enum | READ | Error returned from | | | | | | getattr during | | | | | | readdir. | | filehandle | 19 | nfs_fh4 | READ | The filehandle of | | | | | | this object | | | | | | (primarily for | | | | | | readdir requests). | +-----------------+----+------------+--------+----------------------+ 5.6. Recommended Attributes - Definitions Shepler, et al. Expires September 5, 2007 [Page 84] Internet-Draft NFSv4 Minor Version 1 March 2007 +-------------------+----+----------------+--------+----------------+ | name | # | Data Type | Access | Description | +-------------------+----+----------------+--------+----------------+ | ACL | 12 | nfsace4<> | R/W | The access | | | | | | control list | | | | | | for the | | | | | | object. | | aclsupport | 13 | uint32 | READ | Indicates what | | | | | | types of ACLs | | | | | | are supported | | | | | | on the current | | | | | | file system. | | archive | 14 | bool | R/W | True, if this | | | | | | file has been | | | | | | archived since | | | | | | the time of | | | | | | last | | | | | | modification | | | | | | (deprecated in | | | | | | favor of | | | | | | time_backup). | | cansettime | 15 | bool | READ | True, if the | | | | | | server able to | | | | | | change the | | | | | | times for a | | | | | | file system | | | | | | object as | | | | | | specified in a | | | | | | SETATTR | | | | | | operation. | | case_insensitive | 16 | bool | READ | True, if | | | | | | filename | | | | | | comparisons on | | | | | | this file | | | | | | system are | | | | | | case | | | | | | insensitive. | | case_preserving | 17 | bool | READ | True, if | | | | | | filename case | | | | | | on this file | | | | | | system are | | | | | | preserved. | Shepler, et al. Expires September 5, 2007 [Page 85] Internet-Draft NFSv4 Minor Version 1 March 2007 | chown_restricted | 18 | bool | READ | If TRUE, the | | | | | | server will | | | | | | reject any | | | | | | request to | | | | | | change either | | | | | | the owner or | | | | | | the group | | | | | | associated | | | | | | with a file if | | | | | | the caller is | | | | | | not a | | | | | | privileged | | | | | | user (for | | | | | | example, | | | | | | "root" in UNIX | | | | | | operating | | | | | | environments | | | | | | or in Windows | | | | | | 2000 the "Take | | | | | | Ownership" | | | | | | privilege). | | dacl | 58 | nfsacl41 | R/W | Automatically | | | | | | inheritable | | | | | | access control | | | | | | list used for | | | | | | determining | | | | | | access to file | | | | | | system | | | | | | objects. | | dir_notif_delay | 56 | nfstime4 | READ | notification | | | | | | delays on | | | | | | directory | | | | | | attributes | | dirent_ | 57 | nfstime4 | READ | notification | | notif_delay | | | | delays on | | | | | | child | | | | | | attributes | | fileid | 20 | uint64 | READ | A number | | | | | | uniquely | | | | | | identifying | | | | | | the file | | | | | | within the | | | | | | file system. | Shepler, et al. Expires September 5, 2007 [Page 86] Internet-Draft NFSv4 Minor Version 1 March 2007 | files_avail | 21 | uint64 | READ | File slots | | | | | | available to | | | | | | this user on | | | | | | the file | | | | | | system | | | | | | containing | | | | | | this object - | | | | | | this should be | | | | | | the smallest | | | | | | relevant | | | | | | limit. | | files_free | 22 | uint64 | READ | Free file | | | | | | slots on the | | | | | | file system | | | | | | containing | | | | | | this object - | | | | | | this should be | | | | | | the smallest | | | | | | relevant | | | | | | limit. | | files_total | 23 | uint64 | READ | Total file | | | | | | slots on the | | | | | | file system | | | | | | containing | | | | | | this object. | | fs_absent | 60 | bool | READ | Is current | | | | | | file system | | | | | | present or | | | | | | absent. | | fs_layout_type | 62 | layouttype4<> | READ | Layout types | | | | | | available for | | | | | | the file | | | | | | system. | | fs_locations | 24 | fs_locations | READ | Locations | | | | | | where this | | | | | | file system | | | | | | may be found. | | | | | | If the server | | | | | | returns | | | | | | NFS4ERR_MOVED | | | | | | as an error, | | | | | | this attribute | | | | | | MUST be | | | | | | supported. | | fs_locations_info | 67 | | READ | Full function | | | | | | file system | | | | | | location. | Shepler, et al. Expires September 5, 2007 [Page 87] Internet-Draft NFSv4 Minor Version 1 March 2007 | fs_status | 61 | fs4_status | READ | Generic file | | | | | | system type | | | | | | information. | | hidden | 25 | bool | R/W | True, if the | | | | | | file is | | | | | | considered | | | | | | hidden with | | | | | | respect to the | | | | | | Windows API? | | homogeneous | 26 | bool | READ | True, if this | | | | | | object's file | | | | | | system is | | | | | | homogeneous, | | | | | | i.e. are per | | | | | | file system | | | | | | attributes the | | | | | | same for all | | | | | | file system's | | | | | | objects. | | layout_alignment | 66 | uint32_t | READ | Preferred | | | | | | alignment for | | | | | | layout related | | | | | | I/O. | | layout_blksize | 65 | uint32_t | READ | Preferred | | | | | | block size for | | | | | | layout related | | | | | | I/O. | | layout_hint | 63 | layouthint4 | WRITE | Client | | | | | | specified hint | | | | | | for file | | | | | | layout. | | layout_type | 64 | layouttype4<> | READ | Layout types | | | | | | available for | | | | | | the file. | | maxfilesize | 27 | uint64 | READ | Maximum | | | | | | supported file | | | | | | size for the | | | | | | file system of | | | | | | this object. | | maxlink | 28 | uint32 | READ | Maximum number | | | | | | of links for | | | | | | this object. | | maxname | 29 | uint32 | READ | Maximum | | | | | | filename size | | | | | | supported for | | | | | | this object. | Shepler, et al. Expires September 5, 2007 [Page 88] Internet-Draft NFSv4 Minor Version 1 March 2007 | maxread | 30 | uint64 | READ | Maximum read | | | | | | size supported | | | | | | for this | | | | | | object. | | maxwrite | 31 | uint64 | READ | Maximum write | | | | | | size supported | | | | | | for this | | | | | | object. This | | | | | | attribute | | | | | | SHOULD be | | | | | | supported if | | | | | | the file is | | | | | | writable. | | | | | | Lack of this | | | | | | attribute can | | | | | | lead to the | | | | | | client either | | | | | | wasting | | | | | | bandwidth or | | | | | | not receiving | | | | | | the best | | | | | | performance. | | mdsthreshold | 68 | mdsthreshold4 | READ | Hint to client | | | | | | as to when to | | | | | | write through | | | | | | the pnfs | | | | | | metadata | | | | | | server. | | mimetype | 32 | utf8<> | R/W | MIME body | | | | | | type/subtype | | | | | | of this | | | | | | object. | | mode | 33 | mode4 | R/W | UNIX-style | | | | | | mode including | | | | | | permission | | | | | | bits for this | | | | | | object. | | mode_set_masked | 74 | mode_masked4 | WRITE | Allows setting | | | | | | or resetting a | | | | | | subset of the | | | | | | bits in a | | | | | | UNIX-style | | | | | | mode | Shepler, et al. Expires September 5, 2007 [Page 89] Internet-Draft NFSv4 Minor Version 1 March 2007 | mounted_on_fileid | 55 | uint64 | READ | Like fileid, | | | | | | but if the | | | | | | target | | | | | | filehandle is | | | | | | the root of a | | | | | | file system | | | | | | return the | | | | | | fileid of the | | | | | | underlying | | | | | | directory. | | no_trunc | 34 | bool | READ | True, if a | | | | | | name longer | | | | | | than name_max | | | | | | is used, an | | | | | | error be | | | | | | returned and | | | | | | name is not | | | | | | truncated. | | numlinks | 35 | uint32 | READ | Number of hard | | | | | | links to this | | | | | | object. | | owner | 36 | utf8<> | R/W | The string | | | | | | name of the | | | | | | owner of this | | | | | | object. | | owner_group | 37 | utf8<> | R/W | The string | | | | | | name of the | | | | | | group | | | | | | ownership of | | | | | | this object. | | quota_avail_hard | 38 | uint64 | READ | For definition | | | | | | see "Quota | | | | | | Attributes" | | | | | | section below. | | quota_avail_soft | 39 | uint64 | READ | For definition | | | | | | see "Quota | | | | | | Attributes" | | | | | | section below. | | quota_used | 40 | uint64 | READ | For definition | | | | | | see "Quota | | | | | | Attributes" | | | | | | section below. | Shepler, et al. Expires September 5, 2007 [Page 90] Internet-Draft NFSv4 Minor Version 1 March 2007 | rawdev | 41 | specdata4 | READ | Raw device | | | | | | identifier. | | | | | | UNIX device | | | | | | major/minor | | | | | | node | | | | | | information. | | | | | | If the value | | | | | | of type is not | | | | | | NF4BLK or | | | | | | NF4CHR, the | | | | | | value return | | | | | | SHOULD NOT be | | | | | | considered | | | | | | useful. | | retentevt_get | 71 | retention_get4 | READ | Get the | | | | | | event-based | | | | | | retention | | | | | | duration, and | | | | | | if enabled, | | | | | | the | | | | | | event-based | | | | | | retention | | | | | | begin time of | | | | | | the file | | | | | | object. | | | | | | GETATTR use | | | | | | only. | | retentevt_set | 72 | retention_set4 | WRITE | Set the | | | | | | event-based | | | | | | retention | | | | | | duration, and | | | | | | optionally | | | | | | enable | | | | | | event-based | | | | | | retention on | | | | | | the file | | | | | | object. | | | | | | SETATTR use | | | | | | only. | Shepler, et al. Expires September 5, 2007 [Page 91] Internet-Draft NFSv4 Minor Version 1 March 2007 | retention_get | 69 | retention_get4 | READ | Get the | | | | | | retention | | | | | | duration, and | | | | | | if enabled, | | | | | | the retention | | | | | | begin time of | | | | | | the file | | | | | | object. | | | | | | GETATTR use | | | | | | only. | | retention_hold | 73 | uint64_t | R/W | Get or set | | | | | | administrative | | | | | | retention | | | | | | holds, one | | | | | | hold per bit | | | | | | position. | | retention_set | 70 | retention_set4 | WRITE | Set the | | | | | | retention | | | | | | duration, and | | | | | | optionally | | | | | | enable | | | | | | retention on | | | | | | the file | | | | | | object. | | | | | | SETATTR use | | | | | | only. | | sacl | 59 | nfsacl41 | R/W | Automatically | | | | | | inheritable | | | | | | access control | | | | | | list used for | | | | | | auditing | | | | | | access to | | | | | | files. | | space_avail | 42 | uint64 | READ | Disk space in | | | | | | bytes | | | | | | available to | | | | | | this user on | | | | | | the file | | | | | | system | | | | | | containing | | | | | | this object - | | | | | | this should be | | | | | | the smallest | | | | | | relevant | | | | | | limit. | Shepler, et al. Expires September 5, 2007 [Page 92] Internet-Draft NFSv4 Minor Version 1 March 2007 | space_free | 43 | uint64 | READ | Free disk | | | | | | space in bytes | | | | | | on the file | | | | | | system | | | | | | containing | | | | | | this object - | | | | | | this should be | | | | | | the smallest | | | | | | relevant | | | | | | limit. | | space_total | 44 | uint64 | READ | Total disk | | | | | | space in bytes | | | | | | on the file | | | | | | system | | | | | | containing | | | | | | this object. | | space_used | 45 | uint64 | READ | Number of file | | | | | | system bytes | | | | | | allocated to | | | | | | this object. | | system | 46 | bool | R/W | True, if this | | | | | | file is a | | | | | | "system" file | | | | | | with respect | | | | | | to the Windows | | | | | | API? | | time_access | 47 | nfstime4 | READ | The time of | | | | | | last access to | | | | | | the object by | | | | | | a read that | | | | | | was satisfied | | | | | | by the server. | | time_access_set | 48 | settime4 | WRITE | Set the time | | | | | | of last access | | | | | | to the object. | | | | | | SETATTR use | | | | | | only. | | time_backup | 49 | nfstime4 | R/W | The time of | | | | | | last backup of | | | | | | the object. | Shepler, et al. Expires September 5, 2007 [Page 93] Internet-Draft NFSv4 Minor Version 1 March 2007 | time_create | 50 | nfstime4 | R/W | The time of | | | | | | creation of | | | | | | the object. | | | | | | This attribute | | | | | | does not have | | | | | | any relation | | | | | | to the | | | | | | traditional | | | | | | UNIX file | | | | | | attribute | | | | | | "ctime" or | | | | | | "change time". | | time_delta | 51 | nfstime4 | READ | Smallest | | | | | | useful server | | | | | | time | | | | | | granularity. | | time_metadata | 52 | nfstime4 | READ | The time of | | | | | | last meta-data | | | | | | modification | | | | | | of the object. | | time_modify | 53 | nfstime4 | READ | The time of | | | | | | last | | | | | | modification | | | | | | to the object. | | time_modify_set | 54 | settime4 | WRITE | Set the time | | | | | | of last | | | | | | modification | | | | | | to the object. | | | | | | SETATTR use | | | | | | only. | +-------------------+----+----------------+--------+----------------+ 5.7. Time Access As defined above, the time_access attribute represents the time of last access to the object by a read that was satisfied by the server. The notion of what is an "access" depends on server's operating environment and/or the server's file system semantics. For example, for servers obeying POSIX semantics, time_access would be updated only by the READLINK, READ, and READDIR operations and not any of the operations that modify the content of the object. Of course, setting the corresponding time_access_set attribute is another way to modify the time_access attribute. Whenever the file object resides on a writable file system, the server should make best efforts to record time_access into stable storage. However, to mitigate the performance effects of doing so, and most especially whenever the server is satisfying the read of the Shepler, et al. Expires September 5, 2007 [Page 94] Internet-Draft NFSv4 Minor Version 1 March 2007 object's content from its cache, the server MAY cache access time updates and lazily write them to stable storage. It is also acceptable to give administrators of the server the option to disable time_access updates. 5.8. Interpreting owner and owner_group The recommended attributes "owner" and "owner_group" (and also users and groups within the "acl" attribute) are represented in terms of a UTF-8 string. To avoid a representation that is tied to a particular underlying implementation at the client or server, the use of the UTF-8 string has been chosen. Note that section 6.1 of RFC2624 [27] provides additional rationale. It is expected that the client and server will have their own local representation of owner and owner_group that is used for local storage or presentation to the end user. Therefore, it is expected that when these attributes are transferred between the client and server that the local representation is translated to a syntax of the form "user@ dns_domain". This will allow for a client and server that do not use the same local representation the ability to translate to a common syntax that can be interpreted by both. Similarly, security principals may be represented in different ways by different security mechanisms. Servers normally translate these representations into a common format, generally that used by local storage, to serve as a means of identifying the users corresponding to these security principals. When these local identifiers are translated to the form of the owner attribute, associated with files created by such principals they identify, in a common format, the users associated with each corresponding set of security principals. The translation used to interpret owner and group strings is not specified as part of the protocol. This allows various solutions to be employed. For example, a local translation table may be consulted that maps between a numeric id to the user@dns_domain syntax. A name service may also be used to accomplish the translation. A server may provide a more general service, not limited by any particular translation (which would only translate a limited set of possible strings) by storing the owner and owner_group attributes in local storage without any translation or it may augment a translation method by storing the entire string for attributes for which no translation is available while using the local representation for those cases in which a translation is available. Servers that do not provide support for all possible values of the owner and owner_group attributes, should return an error (NFS4ERR_BADOWNER) when a string is presented that has no translation, as the value to be set for a SETATTR of the owner, Shepler, et al. Expires September 5, 2007 [Page 95] Internet-Draft NFSv4 Minor Version 1 March 2007 owner_group, or acl attributes. When a server does accept an owner or owner_group value as valid on a SETATTR (and similarly for the owner and group strings in an acl), it is promising to return that same string when a corresponding GETATTR is done. Configuration changes and ill-constructed name translations (those that contain aliasing) may make that promise impossible to honor. Servers should make appropriate efforts to avoid a situation in which these attributes have their values changed when no real change to ownership has occurred. The "dns_domain" portion of the owner string is meant to be a DNS domain name. For example, user@ietf.org. Servers should accept as valid a set of users for at least one domain. A server may treat other domains as having no valid translations. A more general service is provided when a server is capable of accepting users for multiple domains, or for all domains, subject to security constraints. In the case where there is no translation available to the client or server, the attribute value must be constructed without the "@". Therefore, the absence of the @ from the owner or owner_group attribute signifies that no translation was available at the sender and that the receiver of the attribute should not use that string as a basis for translation into its own internal format. Even though the attribute value can not be translated, it may still be useful. In the case of a client, the attribute string may be used for local display of ownership. To provide a greater degree of compatibility with previous versions of NFS (i.e. v2 and v3), which identified users and groups by 32-bit unsigned uid's and gid's, owner and group strings that consist of decimal numeric values with no leading zeros can be given a special interpretation by clients and servers which choose to provide such support. The receiver may treat such a user or group string as representing the same user as would be represented by a v2/v3 uid or gid having the corresponding numeric value. A server is not obligated to accept such a string, but may return an NFS4ERR_BADOWNER instead. To avoid this mechanism being used to subvert user and group translation, so that a client might pass all of the owners and groups in numeric form, a server SHOULD return an NFS4ERR_BADOWNER error when there is a valid translation for the user or owner designated in this way. In that case, the client must use the appropriate name@domain string and not the special form for compatibility. The owner string "nobody" may be used to designate an anonymous user, which will be associated with a file created by a security principal that cannot be mapped through normal means to the owner attribute. Shepler, et al. Expires September 5, 2007 [Page 96] Internet-Draft NFSv4 Minor Version 1 March 2007 5.9. Character Case Attributes With respect to the case_insensitive and case_preserving attributes, each UCS-4 character (which UTF-8 encodes) has a "long descriptive name" RFC1345 [28] which may or may not included the word "CAPITAL" or "SMALL". The presence of SMALL or CAPITAL allows an NFS server to implement unambiguous and efficient table driven mappings for case insensitive comparisons, and non-case-preserving storage. For general character handling and internationalization issues, see the section "Internationalization". 5.10. Quota Attributes For the attributes related to file system quotas, the following definitions apply: quota_avail_soft The value in bytes which represents the amount of additional disk space that can be allocated to this file or directory before the user may reasonably be warned. It is understood that this space may be consumed by allocations to other files or directories though there is a rule as to which other files or directories. quota_avail_hard The value in bytes which represent the amount of additional disk space beyond the current allocation that can be allocated to this file or directory before further allocations will be refused. It is understood that this space may be consumed by allocations to other files or directories. quota_used The value in bytes which represent the amount of disc space used by this file or directory and possibly a number of other similar files or directories, where the set of "similar" meets at least the criterion that allocating space to any file or directory in the set will reduce the "quota_avail_hard" of every other file or directory in the set. Note that there may be a number of distinct but overlapping sets of files or directories for which a quota_used value is maintained. E.g. "all files with a given owner", "all files with a given group owner". etc. The server is at liberty to choose any of those sets but should do so in a repeatable way. The rule may be configured per file system or may be "choose the set with the smallest quota". Shepler, et al. Expires September 5, 2007 [Page 97] Internet-Draft NFSv4 Minor Version 1 March 2007 5.11. mounted_on_fileid UNIX-based operating environments connect a file system into the namespace by connecting (mounting) the file system onto the existing file object (the mount point, usually a directory) of an existing file system. When the mount point's parent directory is read via an API like readdir(), the return results are directory entries, each with a component name and a fileid. The fileid of the mount point's directory entry will be different from the fileid that the stat() system call returns. The stat() system call is returning the fileid of the root of the mounted file system, whereas readdir() is returning the fileid stat() would have returned before any file systems were mounted on the mount point. Unlike NFS version 3, NFS version 4 allows a client's LOOKUP request to cross other file systems. The client detects the file system crossing whenever the filehandle argument of LOOKUP has an fsid attribute different from that of the filehandle returned by LOOKUP. A UNIX-based client will consider this a "mount point crossing". UNIX has a legacy scheme for allowing a process to determine its current working directory. This relies on readdir() of a mount point's parent and stat() of the mount point returning fileids as previously described. The mounted_on_fileid attribute corresponds to the fileid that readdir() would have returned as described previously. While the NFS version 4 client could simply fabricate a fileid corresponding to what mounted_on_fileid provides (and if the server does not support mounted_on_fileid, the client has no choice), there is a risk that the client will generate a fileid that conflicts with one that is already assigned to another object in the file system. Instead, if the server can provide the mounted_on_fileid, the potential for client operational problems in this area is eliminated. If the server detects that there is no mounted point at the target file object, then the value for mounted_on_fileid that it returns is the same as that of the fileid attribute. The mounted_on_fileid attribute is RECOMMENDED, so the server SHOULD provide it if possible, and for a UNIX-based server, this is straightforward. Usually, mounted_on_fileid will be requested during a READDIR operation, in which case it is trivial (at least for UNIX- based servers) to return mounted_on_fileid since it is equal to the fileid of a directory entry returned by readdir(). If mounted_on_fileid is requested in a GETATTR operation, the server should obey an invariant that has it returning a value that is equal to the file object's entry in the object's parent directory, i.e. what readdir() would have returned. Some operating environments Shepler, et al. Expires September 5, 2007 [Page 98] Internet-Draft NFSv4 Minor Version 1 March 2007 allow a series of two or more file systems to be mounted onto a single mount point. In this case, for the server to obey the aforementioned invariant, it will need to find the base mount point, and not the intermediate mount points. 5.12. Directory Notification Attributes As described in Section 17.39, the client can request a minimum delay for notifications of changes to attributes, but the server is free ignore what the client requests. The client can determine in advance what notification delays the server will accept by issuing a GETATTR for either or both of two directory notification attributes. When the client calls the GET_DIR_DELEGATION operation and asks^M for attribute change notifications, it should request^M notification delays that are no less than the values in the^M server-provided attributes. 5.12.1. dir_notif_delay The dir_notify_delay attribute is the minimum number of seconds the server will delay before notifying the client of a change to the directory's attributes. 5.12.2. dirent_notif_delay The dirent_notif_delay attribute is the minimum number of seconds the server will delay before notifying the client of a change to a file object that has an entry in the directory. 5.13. PNFS Attributes 5.13.1. fs_layout_type The fs_layout_type attribute (data type layouttype4, see Section 3.2.15) applies to a file system and indicates what layout types are supported by the file system. This attribute is expected be queried when a client encounters a new fsid. This attribute is used by the client to determine if it supports the layout type. 5.13.2. layout_alignment The layout_alignment attribute indicates the preferred alignment for I/O to files on the file system the client has layouts for. Where possible, the client should issue READ and WRITE operations with offsets are whole multiples of the layout_alignment attribute. Shepler, et al. Expires September 5, 2007 [Page 99] Internet-Draft NFSv4 Minor Version 1 March 2007 5.13.3. layout_blksize The layout_blksize attribute indicates the preferred block size for I/O to files on the file system the client has layouts for. Where possible, the client should issue READ operations with a count argument that is a whole multiple of layout_blksize, and WRITE operations with a data argument of size that is a whole multiple of layout_blksize. 5.13.4. layout_hint The layout_hint attribute (data type layouthint4, see Section 3.2.22) may be set on newly created files to influence the metadata server's choice for the file's layout. It is suggested that this attribute is set as one of the initial attributes within the OPEN call. The metadata server may ignore this attribute. This attribute is a sub- set of the layout structure returned by LAYOUTGET. For example, instead of specifying particular devices, this would be used to suggest the stripe width of a file. It is up to the server implementation to determine which fields within the layout it uses. 5.13.5. layout_type This attribute indicates the particular layout type(s) used for a file. This is for informational purposes only. The client needs to use the LAYOUTGET operation in order to get enough information (e.g., specific device information) in order to perform I/O. 5.13.6. mdsthreshold This attribute acts as a hint to the client to help it determine when it is more efficient to issue read and write requests to the metadata server vs. the data server. Two types of thresholds are described: file size thresholds and I/O size thresholds. If a file's size is smaller than the file size threshold, data accesses should be issued to the metadata server. If an I/O is below the I/O size threshold, the I/O should be issued to the metadata server. Each threshold can be specified independently for read and write requests. For either threshold type, a value of 0 indicates no read or write should be issued to the metadata server, while a value of all 1s indicates all reads or writes should be issued to the metadata server. The attribute is available on a per filehandle basis. If the current filehandle refers to a non-pNFS file or directory, the metadata server should return an attribute that is representative of the filehandle's file system. It is suggested that this attribute is queried as part of the OPEN operation. Due to dynamic system changes, the client should not assume that the attribute will remain Shepler, et al. Expires September 5, 2007 [Page 100] Internet-Draft NFSv4 Minor Version 1 March 2007 constant for any specific time period, thus it should be periodically refreshed. 5.14. Retention Attributes Retention is a concept whereby a file object can be placed in an immutable, undeletable, unrenamable state for a fixed or infinite duration of time. Once in this "retained" state, the file cannot be moved out of the state until the duration of retention has been reached. When retention is enabled, retention MUST extend to the data of the file, and the name of file. The server MAY extend retention any other property of the file, including any subset of mandatory, recommended, and named attributes, with the exceptions noted in this section. Servers MAY support or not support retention on any file object type. There are five retention attributes: o retention_get. This attribute is only readable via GETATTR and not setable via SETATTR. The value of the attribute consists of: const RET4_DURATION_INFINITE = 0xffffffffffffffff; struct retention_get4 { uint64_t rg_duration; nfstime4 rg_begin_time<1>; }; The field rg_duration is duration in seconds indicating how long the file will be retained once retention is enabled. The field rg_begin_time is an array of up to one absolute time value. If the array is zero length, no beginning retention time has been established, and retention is not enabled. If rg_duration is equal to RET4_DURATION_INFINITE, the file, once retention is enabled, will be retained for an infinite duration. o retention_set. This attribute corresponds to retention_get. This attribute is only setable via SETATTR and not readable via GETATTR. The value of the attribute consists of: struct retention_set4 { bool rs_enable; uint64_t rs_duration<1>; }; Shepler, et al. Expires September 5, 2007 [Page 101] Internet-Draft NFSv4 Minor Version 1 March 2007 If the client sets rs_enable to TRUE, then it is enabling retention on the file object with the begin time of retention commencing from the server's current time and date. The duration of the retention can also be provided if the rs_duration array is of length one. The duration is time is seconds from the begin time of retention, and if set to RET4_DURATION_INFINITE, the file is to be retained forever. If retention is enabled, with no duration specified in either this SETATTR or a previous SETATTR, the duration defaults to zero seconds. The server MAY restrict the enabling of retention or the duration of retention on the basis of the ACE4_WRITE_RETENTION ACL permission. The enabling of retention does not prevent the enabling of event-based retention nor the modification of the retention_hold attribute. o retentevt_get. This attribute is like retention_get, but refers to event-based retention. The event that triggers event-based retention is not defined by the NFSv4.1 specification. o retentevt_set. This attribute corresponds to retentevt_get, is like retention_set, but refers to event-based retention. When event based retention is set, the file MUST be retained even if non-event-based retention has been set, and the duration of non- event-based retention has been reached. Conversely, when non- event-based retention has been set, the file MUST be retained even the event-based retention has been set, and the duration of event- based retention has been reached. The server MAY restrict the enabling of event-based retention or the duration of event-based retention on the basis of the ACE4_WRITE_RETENTION ACL permission. The enabling of event-based retention does not prevent the enabling of non-event-based retention nor the modification of the retention_hold attribute. o retention_hold. This attribute allows one to 64 administrative holds, one hold per bit on the attribute. If retention_hold is not zero, then the file MUST NOT be deleted, renamed, or modified, even if the duration on enabled event or non-event-based retention has been reached. The server MAY restrict the modification of retention_hold on the basis of the ACE4_WRITE_RETENTION_HOLD ACL permission. The enabling of administration retention holds does not prevent the enabling of event-based or non-event-based retention. Shepler, et al. Expires September 5, 2007 [Page 102] Internet-Draft NFSv4 Minor Version 1 March 2007 6. Access Control Lists Access Control Lists (ACLs) are a file attribute that specify fine grained access control. This chapter covers the "acl", "dacl", "sacl", "aclsupport", "mode", "mode_set_masked" file attributes, and their interactions. 6.1. Goals ACLs and modes represent two well established but different models for specifying permissions. This chapter specifies requirements that attempt to meet the following goals: o If a server supports the mode attribute, it should provide reasonable semantics to clients that only set and retrieve the mode attribute. o If a server supports the ACL attribute, it should provide reasonable semantics to clients that only set and retrieve the ACL attribute. o On servers that support the mode attribute, if the ACL attribute has never been set on an object, via inheritance or explicitly, the behavior should be traditional UNIX-like behavior. o On servers that support the mode attribute, if the ACL attribute has been previously set on an object, either explicitly or via inheritance: * Setting only the mode attribute should effectively control the traditional UNIX-like permissions of read, write, and execute on owner, owner_group, and other. * Setting only the mode attribute should provide reasonable security. For example, setting a mode of 000 should be enough to ensure that future opens for read or write by any principal should fail, regardless of a previously existing or inherited ACL. o This minor version of NFSv4 should not introduce significantly different semantics relating to the mode and ACL attributes, nor should it render invalid any existing implementations. Rather, this chapter provides clarifications based on previous implementations and discussions around them. o If a server supports the ACL attribute, then at any time, the server can provide an ACL attribute when requested. The ACL attribute will describe all permissions on the file object, except Shepler, et al. Expires September 5, 2007 [Page 103] Internet-Draft NFSv4 Minor Version 1 March 2007 for the three high-order bits of the mode attribute (described in Section 6.2.3). The ACL attribute will not conflict with the mode attribute, on servers that support the mode attribute. o If a server supports the mode attribute, then at any time, the server can provide a mode attribute when requested. The mode attribute will not conflict with the ACL attribute, on servers that support the ACL attribute. o When a mode attribute is set on an object, the ACL attribute may need to be modified so as to not conflict with the new mode. In such cases, it is desirable that the ACL keep as much information as possible. This includes information about inheritance, AUDIT and ALARM ACEs, and permissions granted and denied that do not conflict with the new mode. 6.2. File Attributes Discussion 6.2.1. ACL Attribute The NFS version 4 ACL attribute is an array of access control entries (ACEs). Although the client can read and write the ACL attribute, the server is responsible for using the ACL to perform access control. The client can use the OPEN or ACCESS operations to check access without modifying or reading data or metadata. The NFS ACE attribute is defined as follows: typedef uint32_t acetype4; typedef uint32_t aceflag4; typedef uint32_t acemask4; struct nfsace4 { acetype4 type; aceflag4 flag; acemask4 access_mask; utf8str_mixed who; }; To determine if a request succeeds, the server processes each nfsace4 entry in order. Only ACEs which have a "who" that matches the requester are considered. Each ACE is processed until all of the bits of the requester's access have been ALLOWED. Once a bit (see below) has been ALLOWED by an ACCESS_ALLOWED_ACE, it is no longer considered in the processing of later ACEs. If an ACCESS_DENIED_ACE is encountered where the requester's access still has unALLOWED bits in common with the "access_mask" of the ACE, the request is denied. When the ACL is fully processed, if there are bits in the requester's Shepler, et al. Expires September 5, 2007 [Page 104] Internet-Draft NFSv4 Minor Version 1 March 2007 mask that have not been ALLOWED or DENIED, access is denied. Unlike the ALLOW and DENY ACE types, the ALARM and AUDIT ACE types do not affect a requester's access, and instead are for triggering events as a result of a requester's access attempt. Therefore, all AUDIT and ALARM ACEs are processed until end of the ACL. The NFS version 4 ACL model is quite rich. Some server platforms may provide access control functionality that goes beyond the UNIX-style mode attribute, but which is not as rich as the NFS ACL model. So that users can take advantage of this more limited functionality, the server may indicate that it supports ACLs as long as it follows the guidelines for mapping between its ACL model and the NFS version 4 ACL model. The situation is complicated by the fact that a server may have multiple modules that enforce ACLs. For example, the enforcement for NFS version 4 access may be different from the enforcement for local access, and both may be different from the enforcement for access through other protocols such as SMB. So it may be useful for a server to accept an ACL even if not all of its modules are able to support it. The guiding principle in all cases is that the server must not accept ACLs that appear to make the file more secure than it really is. 6.2.1.1. ACE Type The constants used for the type field (acetype4) are as follows: const ACE4_ACCESS_ALLOWED_ACE_TYPE = 0x00000000; const ACE4_ACCESS_DENIED_ACE_TYPE = 0x00000001; const ACE4_SYSTEM_AUDIT_ACE_TYPE = 0x00000002; const ACE4_SYSTEM_ALARM_ACE_TYPE = 0x00000003; +------------------------------+--------------+---------------------+ | Value | Abbreviation | Description | +------------------------------+--------------+---------------------+ | ACE4_ACCESS_ALLOWED_ACE_TYPE | ALLOW | Explicitly grants | | | | the access defined | | | | in acemask4 to the | | | | file or directory. | | ACE4_ACCESS_DENIED_ACE_TYPE | DENY | Explicitly denies | | | | the access defined | | | | in acemask4 to the | | | | file or directory. | Shepler, et al. Expires September 5, 2007 [Page 105] Internet-Draft NFSv4 Minor Version 1 March 2007 | ACE4_SYSTEM_AUDIT_ACE_TYPE | AUDIT | LOG (system | | | | dependent) any | | | | access attempt to a | | | | file or directory | | | | which uses any of | | | | the access methods | | | | specified in | | | | acemask4. | | ACE4_SYSTEM_ALARM_ACE_TYPE | ALARM | Generate a system | | | | ALARM (system | | | | dependent) when any | | | | access attempt is | | | | made to a file or | | | | directory for the | | | | access methods | | | | specified in | | | | acemask4. | +------------------------------+--------------+---------------------+ The "Abbreviation" column denotes how the types will be referred to throughout the rest of this document. 6.2.1.2. The aclsupport Attribute A server need not support all of the above ACE types. The bitmask constants used to represent the above definitions within the aclsupport attribute are as follows: const ACL4_SUPPORT_ALLOW_ACL = 0x00000001; const ACL4_SUPPORT_DENY_ACL = 0x00000002; const ACL4_SUPPORT_AUDIT_ACL = 0x00000004; const ACL4_SUPPORT_ALARM_ACL = 0x00000008; Clients should not attempt to set an ACE unless the server claims support for that ACE type. If the server receives a request to set an ACE that it cannot store, it MUST reject the request with NFS4ERR_ATTRNOTSUPP. If the server receives a request to set an ACE that it can store but cannot enforce, the server SHOULD reject the request with NFS4ERR_ATTRNOTSUPP. Example: suppose a server can enforce NFS ACLs for NFS access but cannot enforce ACLs for local access. If arbitrary processes can run on the server, then the server SHOULD NOT indicate ACL support. On the other hand, if only trusted administrative programs run locally, then the server may indicate ACL support. Shepler, et al. Expires September 5, 2007 [Page 106] Internet-Draft NFSv4 Minor Version 1 March 2007 6.2.1.3. ACE Access Mask The bitmask constants used for the access mask field are as follows: const ACE4_READ_DATA = 0x00000001; const ACE4_LIST_DIRECTORY = 0x00000001; const ACE4_WRITE_DATA = 0x00000002; const ACE4_ADD_FILE = 0x00000002; const ACE4_APPEND_DATA = 0x00000004; const ACE4_ADD_SUBDIRECTORY = 0x00000004; const ACE4_READ_NAMED_ATTRS = 0x00000008; const ACE4_WRITE_NAMED_ATTRS = 0x00000010; const ACE4_EXECUTE = 0x00000020; const ACE4_DELETE_CHILD = 0x00000040; const ACE4_READ_ATTRIBUTES = 0x00000080; const ACE4_WRITE_ATTRIBUTES = 0x00000100; const ACE4_WRITE_RETENTION = 0x00000200; const ACE4_WRITE_RETENTION_HOLD = 0x00000400; const ACE4_DELETE = 0x00010000; const ACE4_READ_ACL = 0x00020000; const ACE4_WRITE_ACL = 0x00040000; const ACE4_WRITE_OWNER = 0x00080000; const ACE4_SYNCHRONIZE = 0x00100000; 6.2.1.3.1. Discussion of Mask Attributes ACE4_READ_DATA Operation(s) affected: READ OPEN Discussion: Permission to read the data of the file. Servers SHOULD allow a user the ability to read the data of the file when only the ACE4_EXECUTE access mask bit is allowed. ACE4_LIST_DIRECTORY Operation(s) affected: READDIR Discussion: Permission to list the contents of a directory. ACE4_WRITE_DATA Operation(s) affected: WRITE OPEN SETATTR of size Shepler, et al. Expires September 5, 2007 [Page 107] Internet-Draft NFSv4 Minor Version 1 March 2007 Discussion: Permission to modify a file's data anywhere in the file's offset range. This includes the ability to write to any arbitrary offset and as a result to grow the file. ACE4_ADD_FILE Operation(s) affected: CREATE OPEN Discussion: Permission to add a new file in a directory. The CREATE operation is affected when nfs_ftype4 is NF4LNK, NF4BLK, NF4CHR, NF4SOCK, or NF4FIFO. (NF4DIR is not listed because it is covered by ACE4_ADD_SUBDIRECTORY.) OPEN is affected when used to create a regular file. ACE4_APPEND_DATA Operation(s) affected: WRITE OPEN SETATTR of size Discussion: The ability to modify a file's data, but only starting at EOF. This allows for the notion of append-only files, by allowing ACE4_APPEND_DATA and denying ACE4_WRITE_DATA to the same user or group. If a file has an ACL such as the one described above and a WRITE request is made for somewhere other than EOF, the server SHOULD return NFS4ERR_ACCESS. ACE4_ADD_SUBDIRECTORY Operation(s) affected: CREATE Discussion: Permission to create a subdirectory in a directory. The CREATE operation is affected when nfs_ftype4 is NF4DIR. ACE4_READ_NAMED_ATTRS Operation(s) affected: OPENATTR Discussion: Permission to read the named attributes of a file or to lookup the named attributes directory. OPENATTR is affected when it is not used to create a named attribute directory. This is when 1.) createdir is TRUE, but a named attribute directory already exists, or 2.) createdir is FALSE. Shepler, et al. Expires September 5, 2007 [Page 108] Internet-Draft NFSv4 Minor Version 1 March 2007 ACE4_WRITE_NAMED_ATTRS Operation(s) affected: OPENATTR Discussion: Permission to write the named attributes of a file or to create a named attribute directory. OPENATTR is affected when it is used to create a named attribute directory. This is when createdir is TRUE and no named attribute directory exists. The ability to check whether or not a named attribute directory exists depends on the ability to look it up, therefore, users also need the ACE4_READ_NAMED_ATTRS permission in order to create a named attribute directory. ACE4_EXECUTE Operation(s) affected: LOOKUP READ OPEN Discussion: Permission to execute a file or traverse/search a directory. Servers SHOULD allow a user the ability to read the data of the file when only the ACE4_EXECUTE access mask bit is allowed. This is because there is no way to execute a file without reading the contents. Though a server may treat ACE4_EXECUTE and ACE4_READ_DATA bits identically when deciding to permit a READ operation, it SHOULD still allow the two bits to be set independently in ACLs, and MUST distinguish between them when replying to ACCESS operations. In particular, servers SHOULD NOT silently turn on one of the two bits when the other is set, as that would make it impossible for the client to correctly enforce the distinction between read and execute permissions. As an example, following a SETATTR of the following ACL: nfsuser:ACE4_EXECUTE:ALLOW A subsequent GETATTR of ACL for that file SHOULD return: nfsuser:ACE4_EXECUTE:ALLOW Rather than: nfsuser:ACE4_EXECUTE/ACE4_READ_DATA:ALLOW ACE4_DELETE_CHILD Operation(s) affected: REMOVE Shepler, et al. Expires September 5, 2007 [Page 109] Internet-Draft NFSv4 Minor Version 1 March 2007 Discussion: Permission to delete a file or directory within a directory. See section "ACE4_DELETE vs. ACE4_DELETE_CHILD" for information on how these two access mask bits interact. ACE4_READ_ATTRIBUTES Operation(s) affected: GETATTR of file system object attributes Discussion: The ability to read basic attributes (non-ACLs) of a file. On a UNIX system, basic attributes can be thought of as the stat level attributes. Allowing this access mask bit would mean the entity can execute "ls -l" and stat. ACE4_WRITE_ATTRIBUTES Operation(s) affected: SETATTR of time_access_set, time_backup, time_create, time_modify_set, mimetype, hidden, system Discussion: Permission to change the times associated with a file or directory to an arbitrary value. Also permission to change the mimetype, hidden and system attributes. A user having ACE4_WRITE_DATA permission, but lacking ACE4_WRITE_ATTRIBUTES must be allowed to implicitly set the times associated with a file. ACE4_WRITE_RETENTION Operation(s) affected: SETATTR of retention_set, retentevt_set. Discussion: Permission to modify the durations of event and non-event-based retention. Also permission to enable event and non-event-based retention. A server MAY map ACE4_WRITE_ATTRIBUTES to ACE_WRITE_RETENTION. ACE4_WRITE_RETENTION_HOLD Operation(s) affected: SETATTR of retention_hold. Discussion: Permission to modify the administration retention holds. A server MAY map ACE4_WRITE_ATTRIBUTES to ACE_WRITE_RETENTION_HOLD. ACE4_DELETE Operation(s) affected: REMOVE Discussion: Permission to delete the file or directory. See section Shepler, et al. Expires September 5, 2007 [Page 110] Internet-Draft NFSv4 Minor Version 1 March 2007 "ACE4_DELETE vs. ACE4_DELETE_CHILD" for information on how these two access mask bits interact. ACE4_READ_ACL Operation(s) affected: GETATTR of acl Discussion: Permission to read the ACL. ACE4_WRITE_ACL Operation(s) affected: SETATTR of acl and mode Discussion: Permission to write the acl and mode attributes. ACE4_WRITE_OWNER Operation(s) affected: SETATTR of owner and owner_group Discussions: Permission to write the owner and owner_group attributes. On UNIX systems, this is the ability to execute chown(). ACE4_SYNCHRONIZE Operation(s) affected: NONE Discussion: Permission to access file locally at the server with synchronized reads and writes. Server implementations need not provide the granularity of control that is implied by this list of masks. For example, POSIX-based systems might not distinguish ACE4_APPEND_DATA (the ability to append to a file) from ACE4_WRITE_DATA (the ability to modify existing contents); both masks would be tied to a single "write" permission. When such a server returns attributes to the client, it would show both ACE4_APPEND_DATA and ACE4_WRITE_DATA if and only if the write permission is enabled. If a server receives a SETATTR request that it cannot accurately implement, it should error in the direction of more restricted access. For example, suppose a server cannot distinguish overwriting data from appending new data, as described in the previous paragraph. If a client submits an ACE where ACE4_APPEND_DATA is set but ACE4_WRITE_DATA is not (or vice versa), the server should reject the request with NFS4ERR_ATTRNOTSUPP. Nonetheless, if the ACE has type DENY, the server may silently turn on the other bit, so that both ACE4_APPEND_DATA and ACE4_WRITE_DATA are denied. Shepler, et al. Expires September 5, 2007 [Page 111] Internet-Draft NFSv4 Minor Version 1 March 2007 6.2.1.3.2. ACE4_DELETE vs. ACE4_DELETE_CHILD Two access mask bits govern the ability to delete a file or directory object: ACE4_DELETE on the object itself, and ACE4_DELETE_CHILD on the object's parent directory. Many systems also consult the "sticky bit" (MODE4_SVTX) and write mode bit on the parent directory when determining whether to allow a file to be deleted. The mode bit for write corresponds to ACE4_WRITE_DATA, which is the same physical bit as ACE4_ADD_FILE. Therefore, ACE4_ADD_FILE can come into play when determining permission to delete. In the algorithm below, the strategy is that ACE4_DELETE and ACE4_DELETE_CHILD take precedence over the sticky bit, and the sticky bit takes precedence over the "write" mode bits (reflected in ACE4_ADD_FILE). Server implementations SHOULD grant or deny permission to delete based on the following algorithm. if ACE4_EXECUTE is denied by the parent directory ACL: deny delete else if ACE4_DELETE is allowed by the target object ACL: allow delete else if ACE4_DELETE_CHILD is allowed by the parent directory ACL: allow delete else if ACE4_DELETE_CHILD is denied by the parent directory ACL: deny delete else if ACE4_ADD_FILE is allowed by the parent directory ACL: if MODE4_SVTX is set for the parent directory: if the principal owns the parent directory OR the principal owns the target object OR ACE4_WRITE_DATA is allowed by the target object ACL: allow delete else: deny delete else: allow delete else: deny delete Shepler, et al. Expires September 5, 2007 [Page 112] Internet-Draft NFSv4 Minor Version 1 March 2007 6.2.1.4. ACE flag The bitmask constants used for the flag field are as follows: const ACE4_FILE_INHERIT_ACE = 0x00000001; const ACE4_DIRECTORY_INHERIT_ACE = 0x00000002; const ACE4_NO_PROPAGATE_INHERIT_ACE = 0x00000004; const ACE4_INHERIT_ONLY_ACE = 0x00000008; const ACE4_SUCCESSFUL_ACCESS_ACE_FLAG = 0x00000010; const ACE4_FAILED_ACCESS_ACE_FLAG = 0x00000020; const ACE4_IDENTIFIER_GROUP = 0x00000040; const ACE4_INHERITED_ACE = 0x00000080; A server need not support any of these flags. If the server supports flags that are similar to, but not exactly the same as, these flags, the implementation may define a mapping between the protocol-defined flags and the implementation-defined flags. Again, the guiding principle is that the file not appear to be more secure than it really is. For example, suppose a client tries to set an ACE with ACE4_FILE_INHERIT_ACE set but not ACE4_DIRECTORY_INHERIT_ACE. If the server does not support any form of ACL inheritance, the server should reject the request with NFS4ERR_ATTRNOTSUPP. If the server supports a single "inherit ACE" flag that applies to both files and directories, the server may reject the request (i.e., requiring the client to set both the file and directory inheritance flags). The server may also accept the request and silently turn on the ACE4_DIRECTORY_INHERIT_ACE flag. 6.2.1.4.1. Discussion of Flag Bits ACE4_FILE_INHERIT_ACE Can be placed on a directory and indicates that this ACE should be added to each new non-directory file created. ACE4_DIRECTORY_INHERIT_ACE Can be placed on a directory and indicates that this ACE should be added to each new directory created. ACE4_INHERIT_ONLY_ACE Can be placed on a directory but does not apply to the directory; ALLOW and DENY ACEs with this bit set do not affect access to the directory, and AUDIT and ALARM ACEs with this bit set do not trigger log or alarm events. Such ACEs only take effect once they are applied (with this bit cleared) to newly created files and directories as specified by the above two flags. Shepler, et al. Expires September 5, 2007 [Page 113] Internet-Draft NFSv4 Minor Version 1 March 2007 ACE4_NO_PROPAGATE_INHERIT_ACE Can be placed on a directory. This flag tells the server that inheritance of this ACE should stop at newly created child directories. ACE4_INHERITED_ACE Indicates that this ACE is inherited from a parent directory. A server that supports automatic inheritance will place this flag on any ACEs inherited from the parent directory when creating a new object. Client applications will use this to perform automatic inheritance. Clients and servers MUST clear this bit in the acl attribute; it may only be used in the dacl and sacl attributes. ACE4_SUCCESSFUL_ACCESS_ACE_FLAG ACE4_FAILED_ACCESS_ACE_FLAG The ACE4_SUCCESSFUL_ACCESS_ACE_FLAG (SUCCESS) and ACE4_FAILED_ACCESS_ACE_FLAG (FAILED) flag bits relate only to ACE4_SYSTEM_AUDIT_ACE_TYPE (AUDIT) and ACE4_SYSTEM_ALARM_ACE_TYPE (ALARM) ACE types. If during the processing of the file's ACL, the server encounters an AUDIT or ALARM ACE that matches the principal attempting the OPEN, the server notes that fact, and the presence, if any, of the SUCCESS and FAILED flags encountered in the AUDIT or ALARM ACE. Once the server completes the ACL processing, it then notes if the operation succeeded or failed. If the operation succeeded, and if the SUCCESS flag was set for a matching AUDIT or ALARM ACE, then the appropriate AUDIT or ALARM event occurs. If the operation failed, and if the FAILED flag was set for the matching AUDIT or ALARM ACE, then the appropriate AUDIT or ALARM event occurs. Either or both of the SUCCESS or FAILED can be set, but if neither is set, the AUDIT or ALARM ACE is not useful. The previously described processing applies to that of the ACCESS operation as well, the difference being that "success" or "failure" does not mean whether ACCESS returns NFS4_OK or not. Success means whether ACCESS returns all requested and supported bits. Failure means whether ACCESS failed to return a bit that was requested and supported. ACE4_IDENTIFIER_GROUP Indicates that the "who" refers to a GROUP as defined under UNIX or a GROUP ACCOUNT as defined under Windows. Clients and servers must ignore the ACE4_IDENTIFIER_GROUP flag on ACEs with a who value equal to one of the special identifiers outlined in Section 6.2.1.5. Shepler, et al. Expires September 5, 2007 [Page 114] Internet-Draft NFSv4 Minor Version 1 March 2007 6.2.1.5. ACE Who The "who" field of an ACE is an identifier that specifies the principal or principals to whom the ACE applies. It may refer to a user or a group, with the flag bit ACE4_IDENTIFIER_GROUP specifying which. There are several special identifiers which need to be understood universally, rather than in the context of a particular DNS domain. Some of these identifiers cannot be understood when an NFS client accesses the server, but have meaning when a local process accesses the file. The ability to display and modify these permissions is permitted over NFS, even if none of the access methods on the server understands the identifiers. +---------------+--------------------------------------------------+ | Who | Description | +---------------+--------------------------------------------------+ | OWNER | The owner of the file | | GROUP | The group associated with the file. | | EVERYONE | The world, including the owner and owning group. | | INTERACTIVE | Accessed from an interactive terminal. | | NETWORK | Accessed via the network. | | DIALUP | Accessed as a dialup user to the server. | | BATCH | Accessed from a batch job. | | ANONYMOUS | Accessed without any authentication. | | AUTHENTICATED | Any authenticated user (opposite of ANONYMOUS) | | SERVICE | Access from a system service. | +---------------+--------------------------------------------------+ Table 7 To avoid conflict, these special identifiers are distinguish by an appended "@" and should appear in the form "xxxx@" (note: no domain name after the "@"). For example: ANONYMOUS@. 6.2.1.5.1. Discussion of EVERYONE@ It is important to note that "EVERYONE@" is not equivalent to the UNIX "other" entity. This is because, by definition, UNIX "other" does not include the owner or owning group of a file. "EVERYONE@" means literally everyone, including the owner or owning group. 6.2.2. dacl and sacl Attributes The dacl and sacl attributes are like the acl attribute, but dacl and sacl each allow only certain types of ACEs. The dacl attribute allows just ALLOW and DENY ACEs. The sacl attribute allows just Shepler, et al. Expires September 5, 2007 [Page 115] Internet-Draft NFSv4 Minor Version 1 March 2007 AUDIT and ALARM ACEs. The dacl and sacl attributes also have improved support for automatic inheritance (see Section 6.4.3.2). The separation of ACE types and inheritance support make dacl and sacl a better choice (over acl) for clients when setting ACEs on a file. 6.2.3. mode Attribute The NFS version 4 mode attribute is based on the UNIX mode bits. The following bits are defined: const MODE4_SUID = 0x800; /* set user id on execution */ const MODE4_SGID = 0x400; /* set group id on execution */ const MODE4_SVTX = 0x200; /* save text even after use */ const MODE4_RUSR = 0x100; /* read permission: owner */ const MODE4_WUSR = 0x080; /* write permission: owner */ const MODE4_XUSR = 0x040; /* execute permission: owner */ const MODE4_RGRP = 0x020; /* read permission: group */ const MODE4_WGRP = 0x010; /* write permission: group */ const MODE4_XGRP = 0x008; /* execute permission: group */ const MODE4_ROTH = 0x004; /* read permission: other */ const MODE4_WOTH = 0x002; /* write permission: other */ const MODE4_XOTH = 0x001; /* execute permission: other */ Bits MODE4_RUSR, MODE4_WUSR, and MODE4_XUSR apply to the principal identified in the owner attribute. Bits MODE4_RGRP, MODE4_WGRP, and MODE4_XGRP apply to principals identified in the owner_group attribute but who are not identified in the owner attribute. Bits MODE4_ROTH, MODE4_WOTH, MODE4_XOTH apply to any principal that does not match that in the owner attribute, and does not have a group matching that of the owner_group attribute. Bits within the mode other than those specified above are not defined by this protocol. A server MUST NOT return bits other than those defined above in a GETATTR or READDIR operation, and it MUST return NFS4ERR_INVAL if bits other than those defined above are set in a SETATTR, CREATE, or OPEN operation. 6.2.4. mode_set_masked Attribute The mode_set_masked attribute is a write-only attribute that allows individual bits in the mode attribute to be set or reset, without changing others. It allows, for example, the bits MODE4_SUID, MODE4_SGID, and MODE4_SVTX to be modified while leaving unmodified any of the nine low-order mode bits devoted to permissions. The mode_set_masked attribute consists of two words each in the form of a mode4. The first consists of the value to be applied to the Shepler, et al. Expires September 5, 2007 [Page 116] Internet-Draft NFSv4 Minor Version 1 March 2007 current mode value and the second is a mask. Only bits set to one in the mask word are changed (set or reset) in the file's mode. All other bits in the mode remain unchanged. Bits in the first word that correspond to bits which are zero in the mask are ignored, except that undefined bits are checked for validity and can result in NFSERR_INVAL as described below. The mode_set_masked attribute is only valid in a SETATTR operation. If it is used in a CREATE or OPEN operation, the server MUST return NFS4ERR_INVAL. Bits not defined as valid in the mode attribute are not valid in either word of the mode_set_masked attribute. The server MUST return NFS4ERR_INVAL if any of those are on in a SETATTR. If the mode and mode_set_masked attributes are both specified in the same SETATTR, the server MUST also return NFS4ERR_INVAL. 6.3. Common Methods The requirements in this section will be referred to in future sections, especially Section 6.4. 6.3.1. Interpreting an ACL 6.3.1.1. Server Considerations The server uses the algorithm described in Section 6.2.1 to determine whether an ACL allows access to an object. However, the ACL may not be the sole determiner of access. For example: o In the case of a file system exported as read-only, the server may deny write permissions even though an object's ACL grants it. o Server implementations MAY grant ACE4_WRITE_ACL and ACE4_READ_ACL permissions in order to prevent the owner from getting into the situation where they can't ever modify the ACL. o All servers will allow a user the ability to read the data of the file when only the execute permission is granted (i.e. If the ACL denies the user the ACE4_READ_DATA access and allows the user ACE4_EXECUTE, the server will allow the user to read the data of the file). o Many servers have the notion of owner-override in which the owner of the object is allowed to override accesses that are denied by the ACL. This may be helpful, for example, to allow users continued access to open files on which the permissions have changed. Shepler, et al. Expires September 5, 2007 [Page 117] Internet-Draft NFSv4 Minor Version 1 March 2007 6.3.1.2. Client Considerations Clients SHOULD NOT do their own access checks based on their interpretation the ACL, but rather use the OPEN and ACCESS operations to do access checks. This allows the client to act on the results of having the server determine whether or not access should be granted based on its interpretation of the ACL. Clients must be aware of situations in which an object's ACL will define a certain access even though the server will not enforce it. In general, but especially in these situations, the client needs to do its part in the enforcement of access as defined by the ACL. To do this, the client MAY issue the appropriate ACCESS operation prior to servicing the request of the user or application in order to determine whether the user or application should be granted the access requested. For examples in which the ACL may define accesses that the server doesn't enforce see Section 6.3.1.1. 6.3.2. Computing a Mode Attribute from an ACL The following method can be used to calculate the MODE4_R*, MODE4_W* and MODE4_X* bits of a mode attribute, based upon an ACL. 1. To determine MODE4_ROTH, MODE4_WOTH, and MODE4_XOTH: 1. If the special identifier EVERYONE@ is granted ACE4_READ_DATA, then the bit MODE4_ROTH SHOULD be set. Otherwise, MODE4_ROTH SHOULD NOT be set. 2. If the special identifier EVERYONE@ is granted ACE4_WRITE_DATA or ACE4_APPEND_DATA, then the bit MODE4_WOTH SHOULD be set. Otherwise, MODE4_WOTH SHOULD NOT be set. 3. If the special identifier EVERYONE@ is granted ACE4_EXECUTE, then the bit MODE4_XOTH SHOULD be set. Otherwise, MODE4_XOTH SHOULD NOT be set. 2. To determine MODE4_RGRP, MODE4_WGRP, and MODE4_XGRP, note that the EVERYONE@ special identifier SHOULD be taken into account. In other words, when determining if the GROUP@ special identifier is granted a permission, ACEs with the identifier EVERYONE@ should take effect just as ACEs with the special identifier GROUP@ would. 1. If the special identifier GROUP@ is granted ACE4_READ_DATA, then the bit MODE4_RGRP SHOULD be set. Otherwise, MODE4_RGRP SHOULD NOT be set. Shepler, et al. Expires September 5, 2007 [Page 118] Internet-Draft NFSv4 Minor Version 1 March 2007 2. If the special identifier GROUP@ is granted ACE4_WRITE_DATA or ACE4_APPEND_DATA, then the bit MODE4_WGRP SHOULD be set. Otherwise, MODE4_WGRP SHOULD NOT be set. 3. If the special identifier GROUP@ is granted ACE4_EXECUTE, then the bit MODE4_XGRP SHOULD be set. Otherwise, MODE4_XGRP SHOULD NOT be set. 3. To determine MODE4_RUSR, MODE4_WUSR, and MODE4_XUSR, note that the EVERYONE@ special identifier SHOULD be taken into account. In other words, when determining if the OWNER@ special identifier is granted a permission, ACEs with the identifier EVERYONE@ should take effect just as ACEs with the special identifer OWNER@ would. 1. If the special identifier OWNER@ is granted ACE4_READ_DATA, then the bit MODE4_RUSR SHOULD be set. Otherwise, MODE4_RUSR SHOULD NOT be set. 2. If the special identifier OWNER@ is granted ACE4_WRITE_DATA or ACE4_APPEND_DATA, then the bit MODE4_WUSR SHOULD be set. Otherwise, MODE4_WUSR SHOULD NOT be set. 3. If the special identifier OWNER@ is granted ACE4_EXECUTE, then the bit MODE4_XUSR SHOULD be set. Otherwise, MODE4_XUSR SHOULD NOT be set. 6.3.2.1. Discussion The nine low-order mode bits (MODE4_R*, MODE4_W*, MODE4_X*) correspond to ACE4_READ_DATA, ACE4_WRITE_DATA/ACE4_APPEND_DATA, and ACE4_EXECUTE for OWNER@, GROUP@, and EVERYONE@. On some implementations, mode bits may represent a superset of these permissions, e.g. if a specific user is granted ACE4_WRITE_DATA, then MODE4_WGRP will be set, even though the file's owner_group is not granted ACE4_WRITE_DATA. Server implementations are discouraged from doing this, as experience has shown that this is confusing and annoying to end users. The specifications above also discourage this practice to enforce the semantic that setting the mode attribute effectively specifies read, write, and execute for owner, group, and other. 6.4. Requirements The server that supports both mode and ACL must take care to synchronize the MODE4_*USR, MODE4_*GRP, and MODE4_*OTH bits with the ACEs which have respective who fields of "OWNER@", "GROUP@", and Shepler, et al. Expires September 5, 2007 [Page 119] Internet-Draft NFSv4 Minor Version 1 March 2007 "EVERYONE@" so that the client can see semantically equivalent access permissions exist whether the client asks for owner, owner_group and mode attributes, or for just the ACL. In this section, much is made of the methods in Section 6.3.2. Many requirements refer to this section. But note that the methods have behaviors specified with "SHOULD". This is intentional, to avoid invalidating existing implementations that compute the mode according to the withdrawn POSIX ACL draft (1003.1e draft 17), rather than by actual permissions on owner, group, and other. 6.4.1. Setting the mode and/or ACL Attributes 6.4.1.1. Setting mode and not ACL When any mode permission bits are subject to change, either because the mode attribute was set or because the mode_set_masked attribute was set and the mask included one or more bits from the low-order nine mode bits that control permissions, and the ACL attribute is not explicitly set, the ACL attribute must be modified in accordance with the updated value of the permissions bits within the mode. This must happen even if the value of the permission bits within the mode is the same after the mode is set as before. In cases in which the permissions bits are subject to change, the ACL attribute MUST be modified such that the mode computed via the method in Section 6.3.2 yields the low-order nine bits (MODE4_R*, MODE4_W*, MODE4_X*) of the mode attribute as modified by the attribute change. The ACL SHOULD also be modified such that: 1. If MODE4_RGRP is not set, entities explicitly listed in the ACL other than OWNER@ and EVERYONE@ SHOULD NOT be granted ACE4_READ_DATA. 2. If MODE4_WGRP is not set, entities explicitly listed in the ACL other than OWNER@ and EVERYONE@ SHOULD NOT be granted ACE4_WRITE_DATA or ACE4_APPEND_DATA. 3. If MODE4_XGRP is not set, entities explicitly listed in the ACL other than OWNER@ and EVERYONE@ SHOULD NOT be granted ACE4_EXECUTE. Access mask bits other those listed above, appearing in ALLOW ACEs, MAY also be disabled. Note that ACEs with the flag ACE4_INHERIT_ONLY_ACE set do not affect the permissions of the ACL itself, nor do ACEs of the type AUDIT and ALARM. As such, it is desirable to leave these ACEs unmodified when Shepler, et al. Expires September 5, 2007 [Page 120] Internet-Draft NFSv4 Minor Version 1 March 2007 modifying the ACL attribute. Also note that the requirement may be met by discarding the ACL, in favor of an ACL that represents the mode and only the mode. This is permitted, but it is preferable for a server to preserve as much of the ACL as possible without violating the above requirements. Discarding the ACL makes it effectively impossible for a file created with a mode attribute to inherit an ACL (see Section 6.4.3). 6.4.1.2. Setting ACL and not mode When setting an ACL attribute and not setting the mode or mode_set_masked attributes, the permission bits of the mode need to be derived from the ACL. In this case, the ACL attribute SHOULD be set as given. The nine low-order bits of the mode attribute (MODE4_R*, MODE4_W*, MODE4_X*) MUST be modified to match the result of the method Section 6.3.2. The three high-order bits of the mode (MODE4_SUID, MODE4_SGID, MODE4_SVTX) SHOULD remain unchanged. 6.4.1.3. Setting both ACL and mode When setting both the mode (includes use of either the mode attribute or the mode_set_masked attribute) and the ACL attribute in the same operation, the attributes MUST be applied in this order: mode (or mode_set_masked), then ACL. The mode-related attribute is set as given, then the ACL attribute is set as given, possibly changing the final mode, as described above in Section 6.4.1.2. 6.4.2. Retrieving the mode and/or ACL Attributes This section applies only to servers that support both the mode and the ACL attribute. Some server implementations may have a concept of "objects without ACLs", meaning that all permissions are granted and denied according to the mode attribute, and that no ACL attribute is stored for that object. If an ACL attribute is requested of such a server, the server SHOULD return an ACL that does not conflict with the mode; that is to say, the ACL returned SHOULD represent the nine low-order bits of the mode attribute (MODE4_R*, MODE4_W*, MODE4_X*) as described in Section 6.3.2. For other server implementations, the ACL attribute is always present for every object. Such servers SHOULD store at least the three high- order bits of the mode attribute (MODE4_SUID, MODE4_SGID, MODE4_SVTX). The server SHOULD return a mode attribute if one is requested, and the low-order nine bits of the mode (MODE4_R*, MODE4_W*, MODE4_X*) MUST match the result of applying the method in Shepler, et al. Expires September 5, 2007 [Page 121] Internet-Draft NFSv4 Minor Version 1 March 2007 Section 6.3.2 to the ACL attribute. 6.4.3. Creating New Objects If a server supports the ACL attribute, it may use the ACL attribute on the parent directory to compute an initial ACL attribute for a newly created object. This will be referred to as the inherited ACL within this section. The act of adding one or more ACEs to the inherited ACL that are based upon ACEs in the parent directory's ACL will be referred to as inheriting an ACE within this section. Implementors should standardize on what the behavior of CREATE and OPEN must be depending on the presence or absence of the mode and ACL attributes. 1. If just mode is given: In this case, inheritance SHOULD take place, but the mode MUST be applied to the inherited ACL as described in Section 6.4.1.1, thereby modifying the ACL. 2. If just ACL is given: In this case, inheritance SHOULD NOT take place, and the ACL as defined in the CREATE or OPEN will be set without modification, and the mode modified as in Section 6.4.1.2 3. If both mode and ACL are given: In this case, inheritance SHOULD NOT take place, and both attributes will be set as described in Section 6.4.1.3. 4. If neither mode nor ACL are given: In the case where an object is being created without any initial attributes at all, e.g. an OPEN operation with an opentype4 of OPEN4_CREATE and a createmode4 of EXCLUSIVE4, inheritance SHOULD NOT take place. Instead, the server SHOULD set permissions to deny all access to the newly created object. It is expected that the appropriate client will set the desired attributes in a subsequent SETATTR operation, and the server SHOULD allow that operation to succeed, regardless of what permissions the object is created with. For example, an empty ACL denies all permissions, but the server should allow the owner's SETATTR to succeed even though WRITE_ACL is implicitly denied. In other cases, inheritance SHOULD take place, and no modifications to the ACL will happen. The mode attribute, if Shepler, et al. Expires September 5, 2007 [Page 122] Internet-Draft NFSv4 Minor Version 1 March 2007 supported, MUST be as computed in Section 6.3.2, with the MODE4_SUID, MODE4_SGID and MODE4_SVTX bits clear. It is worth noting that if no inheritable ACEs exist on the parent directory, the file will be created with an empty ACL, thus granting no access. 6.4.3.1. The Inherited ACL If the object being created is not a directory, the inherited ACL SHOULD NOT inherit ACEs from the parent directory ACL unless the ACE4_FILE_INHERIT_FLAG is set. If the object being created is a directory, the inherited ACL should inherit all inheritable ACEs from the parent directory, those that have ACE4_FILE_INHERIT_ACE or ACE4_DIRECTORY_INHERIT_ACE flag set. If the inheritable ACE has ACE4_FILE_INHERIT_ACE set, but ACE4_DIRECTORY_INHERIT_ACE is clear, the inherited ACE on the newly created directory MUST have the ACE4_INHERIT_ONLY_ACE flag set to prevent the directory from being affected by ACEs meant for non- directories. If when a new directory is created and it inherits ACEs from its parent, for each inheritable ACE which affects the directory's permissions, a server MAY create two ACEs on the directory being created; one effective and one which is only inheritable (i.e. has ACE4_INHERIT_ONLY_ACE flag set). This gives the user and the server, in the cases which it must mask certain permissions upon creation, the ability to modify the effective permissions without modifying the ACE which is to be inherited to the new directory's children. When a newly created object is created with attributes, and those attributes contain an ACL attribute and/or a mode attribute, the server MUST apply those attributes to the newly created object, as described in Section 6.4.1. 6.4.3.2. Automatic Inheritance Unlike the acl attribute, the sacl and dacl (see Section 6.2.2) attributes both have an additional flag field. The flag field applies to the entire sacl or dacl; three flag values are defined const ACL4_AUTO_INHERIT = 0x00000001; const ACL4_PROTECTED = 0x00000002; const ACL4_DEFAULTED = 0x00000004; and all other bits must be cleared. The ACE4_INHERITED_ACE flag may be set in the ACEs of the sacl or dacl (whereas it must always be cleared in the acl). Shepler, et al. Expires September 5, 2007 [Page 123] Internet-Draft NFSv4 Minor Version 1 March 2007 Together these features allow a server to support automatic inheritance, which we now explain in more detail. Inheritable ACEs are normally inherited by child objects only at the time that the child objects are created; later modifications to inheritable ACEs do not result in modifications to inherited ACEs on descendents. However, the dacl and sacl provide an optional mechanism which allows a client application to propagate changes to inheritable ACEs to an entire directory hierarchy. A server that supports this performs inheritance at object creation time in the normal way, but also sets the ACE4_INHERITED_ACE flag on any inherited ACEs as they are added to the new object. A client application such as an ACL editor may then propagate changes to inheritable ACEs on a directory by recursively traversing that directory's descendants and modifying each ACL encountered to remove any ACEs with the ACE4_INHERITED_ACE flag and to replace them by the new inheritable ACEs (also with the ACE4_INHERITED_ACE flag set). It uses the existing ACE inheritance flags in the obvious way to decide which ACEs to propagate. (Note that it may encounter further inheritable ACEs when descending the directory hierarchy, and that those will also need to be taken into account when propagating inheritable ACEs to further descendants.) The reach of this propagation may be limited in two ways: first, automatic inheritance is not performed from any directory ACL that has the ACL4_AUTO_INHERIT flag cleared; and second, automatic inheritance stops wherever an ACL with the ACL4_PROTECTED flag is set, preventing modification of that ACL and also (if the ACL is set on a directory) of the ACL on any of the object's descendants. This propagation is performed independently for the sacl and the dacl attributes; thus the ACL4_AUTO_INHERIT and ACL4_PROTECTED flags may be independently set for the sacl and the dacl, and propagation of one type of acl may continue down a hierarchy even where propagation of the other acl has stopped. New objects should be created with a dacl and a sacl that both have the ACL4_PROTECTED flag cleared and the ACL4_AUTO_INHERIT flag set to the same value as that on, respectively, the sacl or dacl of the parent object. Both the dacl and sacl attributes are RECOMMENDED, and a server may support one without supporting the other. Shepler, et al. Expires September 5, 2007 [Page 124] Internet-Draft NFSv4 Minor Version 1 March 2007 A server that supports both the old acl attribute and one or both of the new dacl or sacl attributes must do so in such a way as to keep all three attributes consistent with each other. Thus the ACEs reported in the acl attribute should be the union of the ACEs reported in the dacl and sacl attributes, except that the ACE4_INHERITED_ACE flag must be cleared from the ACEs in the acl. And of course a client that queries only the acl will be unable to determine the values of the sacl or dacl flag fields. When a client performs a SETATTR for the acl attribute, the server SHOULD set the ACL4_PROTECTED flag to true on both the sacl and the dacl. By using the acl attribute, as opposed to the dacl or sacl attributes, the client signals that it may not understand automatic inheritance, and thus cannot be trusted to set an ACL for which automatic inheritance would make sense. When a client application queries an ACL, modifies it, and sets it again, it should leave any ACEs marked with ACE4_INHERITED_ACE unchanged, in their original order, at the end of the ACL. If the application is unable to do this, it should set the ACL4_PROTECTED flag. This behavior is not enforced by servers, but violations of this rule may lead to unexpected results when applications perform automatic inheritance. If a server also supports the mode attribute, it SHOULD set the mode in such a way that leaves inherited ACEs unchanged, in their original order, at the end of the ACL. If it is unable to do so, it SHOULD set the ACL4_PROTECTED flag on the file's dacl. Finally, in the case where the request that creates a new file or directory does not also set permissions for that file or directory, and there are also no ACEs to inherit from the parent's directory, then the server's choice of ACL for the new object is implementation- dependent. In this case, the server SHOULD set the ACL4_DEFAULTED flag on the ACL it chooses for the new object. An application performing automatic inheritance takes the ACL4_DEFAULTED flag as a sign that the ACL should be completely replaced by one generated using the automatic inheritance rules. 7. Single-server Name Space This chapter describes the NFSv4 single-server name space. Single- server namespaces may be presented directly to clients, or they may be used as a basis to form larger multi-server namespaces (e.g. site- wide or organization-wide) to be presented to clients, as described in Section 10. Shepler, et al. Expires September 5, 2007 [Page 125] Internet-Draft NFSv4 Minor Version 1 March 2007 7.1. Server Exports On a UNIX server, the name space describes all the files reachable by pathnames under the root directory or "/". On a Windows NT server the name space constitutes all the files on disks named by mapped disk letters. NFS server administrators rarely make the entire server's file system name space available to NFS clients. More often portions of the name space are made available via an "export" feature. In previous versions of the NFS protocol, the root filehandle for each export is obtained through the MOUNT protocol; the client sends a string that identifies the export of name space and the server returns the root filehandle for it. The MOUNT protocol supports an EXPORTS procedure that will enumerate the server's exports. 7.2. Browsing Exports The NFS version 4 protocol provides a root filehandle that clients can use to obtain filehandles for the exports of a particular server, via a series of LOOKUP operations within a COMPOUND, to traverse a path. A common user experience is to use a graphical user interface (perhaps a file "Open" dialog window) to find a file via progressive browsing through a directory tree. The client must be able to move from one export to another export via single-component, progressive LOOKUP operations. This style of browsing is not well supported by the NFS version 2 and 3 protocols. The client expects all LOOKUP operations to remain within a single server file system. For example, the device attribute will not change. This prevents a client from taking name space paths that span exports. An automounter on the client can obtain a snapshot of the server's name space using the EXPORTS procedure of the MOUNT protocol. If it understands the server's pathname syntax, it can create an image of the server's name space on the client. The parts of the name space that are not exported by the server are filled in with a "pseudo file system" that allows the user to browse from one mounted file system to another. There is a drawback to this representation of the server's name space on the client: it is static. If the server administrator adds a new export the client will be unaware of it. 7.3. Server Pseudo File System NFS version 4 servers avoid this name space inconsistency by presenting all the exports for a given server within the framework of a single namespace, for that server. An NFS version 4 client uses LOOKUP and READDIR operations to browse seamlessly from one export to Shepler, et al. Expires September 5, 2007 [Page 126] Internet-Draft NFSv4 Minor Version 1 March 2007 another. Portions of the server name space that are not exported are bridged via a "pseudo file system" that provides a view of exported directories only. A pseudo file system has a unique fsid and behaves like a normal, read only file system. Based on the construction of the server's name space, it is possible that multiple pseudo file systems may exist. For example, /a pseudo file system /a/b real file system /a/b/c pseudo file system /a/b/c/d real file system Each of the pseudo file systems are considered separate entities and therefore will have its own unique fsid. 7.4. Multiple Roots The DOS and Windows operating environments are sometimes described as having "multiple roots". File Systems are commonly represented as disk letters. MacOS represents file systems as top level names. NFS version 4 servers for these platforms can construct a pseudo file system above these root names so that disk letters or volume names are simply directory names in the pseudo root. 7.5. Filehandle Volatility The nature of the server's pseudo file system is that it is a logical representation of file system(s) available from the server. Therefore, the pseudo file system is most likely constructed dynamically when the server is first instantiated. It is expected that the pseudo file system may not have an on disk counterpart from which persistent filehandles could be constructed. Even though it is preferable that the server provide persistent filehandles for the pseudo file system, the NFS client should expect that pseudo file system filehandles are volatile. This can be confirmed by checking the associated "fh_expire_type" attribute for those filehandles in question. If the filehandles are volatile, the NFS client must be prepared to recover a filehandle value (e.g. with a series of LOOKUP operations) when receiving an error of NFS4ERR_FHEXPIRED. 7.6. Exported Root If the server's root file system is exported, one might conclude that a pseudo-file system is unneeded. This not necessarily so. Assume the following file systems on a server: Shepler, et al. Expires September 5, 2007 [Page 127] Internet-Draft NFSv4 Minor Version 1 March 2007 / disk1 (exported) /a disk2 (not exported) /a/b disk3 (exported) Because disk2 is not exported, disk3 cannot be reached with simple LOOKUPs. The server must bridge the gap with a pseudo-file system. 7.7. Mount Point Crossing The server file system environment may be constructed in such a way that one file system contains a directory which is 'covered' or mounted upon by a second file system. For example: /a/b (file system 1) /a/b/c/d (file system 2) The pseudo file system for this server may be constructed to look like: / (place holder/not exported) /a/b (file system 1) /a/b/c/d (file system 2) It is the server's responsibility to present the pseudo file system that is complete to the client. If the client sends a lookup request for the path "/a/b/c/d", the server's response is the filehandle of the file system "/a/b/c/d". In previous versions of the NFS protocol, the server would respond with the filehandle of directory "/a/b/c/d" within the file system "/a/b". The NFS client will be able to determine if it crosses a server mount point by a change in the value of the "fsid" attribute. 7.8. Security Policy and Name Space Presentation The application of the server's security policy needs to be carefully considered by the implementor. One may choose to limit the viewability of portions of the pseudo file system based on the server's perception of the client's ability to authenticate itself properly. However, with the support of multiple security mechanisms and the ability to negotiate the appropriate use of these mechanisms, the server is unable to properly determine if a client will be able to authenticate itself. If, based on its policies, the server chooses to limit the contents of the pseudo file system, the server may effectively hide file systems from a client that may otherwise have legitimate access. As suggested practice, the server should apply the security policy of Shepler, et al. Expires September 5, 2007 [Page 128] Internet-Draft NFSv4 Minor Version 1 March 2007 a shared resource in the server's namespace to the components of the resource's ancestors. For example: / /a/b /a/b/c The /a/b/c directory is a real file system and is the shared resource. The security policy for /a/b/c is Kerberos with integrity. The server should apply the same security policy to /, /a, and /a/b. This allows for the extension of the protection of the server's namespace to the ancestors of the real shared resource. For the case of the use of multiple, disjoint security mechanisms in the server's resources, the security for a particular object in the server's namespace should be the union of all security mechanisms of all direct descendants. 8. File Locking and Share Reservations Integrating locking into the NFS protocol necessarily causes it to be stateful. With the inclusion of such features as share reservations, file and directory delegations, recallable layouts, and support for mandatory record locking the protocol becomes substantially more dependent on state than the traditional combination of NFS and NLM [XNFS]. There are three components to making this state manageable: o Clear division between client and server o Ability to reliably detect inconsistency in state between client and server o Simple and robust recovery mechanisms In this model, the server owns the state information. The client requests changes in locks and the server responds with the changes made. Non-client-initiated changes in locking state are infrequent and the client receives prompt notification of them and can adjust its view of the locking state to reflect the server's changes. To support Win32 share reservations it is necessary to provide operations which atomically OPEN or CREATE files. Having a separate share/unshare operation would not allow correct implementation of the Win32 OpenFile API. In order to correctly implement share semantics, the previous NFS protocol mechanisms used when a file is opened or created (LOOKUP, CREATE, ACCESS) need to be replaced. The NFS version 4.1 protocol defines OPEN operation which looks up or creates Shepler, et al. Expires September 5, 2007 [Page 129] Internet-Draft NFSv4 Minor Version 1 March 2007 a file and establishes locking state on the server. 8.1. Locking It is assumed that manipulating a lock is rare when compared to READ and WRITE operations. It is also assumed that crashes and network partitions are relatively rare. Therefore it is important that the READ and WRITE operations have a lightweight mechanism to indicate if they possess a held lock. A lock request contains the heavyweight information required to establish a lock and uniquely define the lock owner. The following sections describe the transition from the heavyweight information to the eventual lightweight stateid used for most client and server locking interactions. 8.1.1. Client and Session ID A client must establish a client ID (see Section 2.4) and then one or more sessionids (see Section 2.10) before performing any operations to open, lock, or delegate a file object. The sessionid services as a shorthand referral to an NFSv4.1 client. 8.1.2. State-owner Definition When opening a file or requesting a record lock, the client must specify an identifier which represents the owner of the requested lock. This identifier is in the form of a state-owner, represented in the protocol by a state_owner4, a variable-length opaque array which, when concatenated with the current client ID uniquely defines the owner of lock managed by the client. This may be a thread id, process id, or other unique value. Owners of opens and owners of record locks are separate entities and remain separate even if the same opaque arrays are used to designate owners of each. The protocol distinguishes between open-owners (represented by open_owner4 structures) and lock-owners (represented by lock_owner4 structures). Each open is associated with a specific open-owner while each record lock is associated with a lock-owner and an open-owner, the latter being the open-owner associated with the open file under which the LOCK operation was done. Delegations and layouts, on the other hand, are not associated with a specific owner but are associated the client as a whole. Shepler, et al. Expires September 5, 2007 [Page 130] Internet-Draft NFSv4 Minor Version 1 March 2007 8.1.3. Stateid Definition When the server grants a lock of any type (including opens, record locks, delegations, and layouts) it responds with a unique stateid, that represents a set of locks (often a single lock) for the same file, of the same type, and sharing the same ownership characteristics. Thus opens of the same file by different open- owners each have an identifying stateid. Similarly, each set of record locks on a file owned by a specific lock-owner and gotten via an open for a specific open-owner, has its own identifying stateid. Delegations and layouts also have associated stateids by which they may be referenced. The stateid is used as a shorthand reference to a lock or set of locks and given a stateid the client can determine the associated state-owner or state-owners (in the case of an open-owner/ lock-owner pair) and the associated filehandle. When stateids are used the current filehandle must be the one associated with that stateid. The server may assign stateids independently for different clients and a stateid with the same bit pattern for one client may designate an entirely different set of locks for a different client. The stateid is always interpreted with respect to the client ID associated with the current session. Stateids apply to all sessions associated with the given client ID and the client may use a stateid obtained from one session on another session associated with the same client ID. 8.1.3.1. Stateid Structure Stateids are divided into two fields, a 96-bit "other" field identifying the specific set of locks and a 32-bit "seqid" sequence value. Except in the case of special stateids, to be discussed below, the purpose of the sequence value within NFSv4.1 is to allow the server to communicate to the client the order in which operations that modified locking state associated with a stateid have been processed. In the case of stateids associated with opens, i.e. the stateids returned by OPEN (the state for the open, rather than that for the delegation), OPEN_DOWNGRADE, or CLOSE, the server MUST provide an "seqid" value starting at one for the first use of a given "other" value and incremented by one with each subsequent operation returning a stateid. In the case of other sorts of stateids (i.e. stateids associated with record locks and delegations), the server MAY provide an incrementing sequence value on successive stateids returned with same identifying field, or it may return the value zero. If it does return a non-zero Shepler, et al. Expires September 5, 2007 [Page 131] Internet-Draft NFSv4 Minor Version 1 March 2007 "seqid" value it MUST start at one and be incremented by one with each subsequent operation returning a stateid with same "other" value, just as is done with open state. The client when using a stateid as a parameter to an operation, must, except in the case of a special stateid, set the sequence value to zero. If the value is non-zero, the server MUST return the error NFS4ERR_BAD_STATEID. 8.1.3.2. Special Stateids Stateid values whose "other" field is either all zeros or all ones are reserved. They may not be assigned by the server but have special meanings defined by the protocol. The particular meaning depends on whether the "other" field is all zeros or all ones and the specific value of the "seqid" field. The following combinations of "other" and "seqid" are defined in NFSv4.1: o When "other" and "seqid" are both zero, the stateid is treated as a special anonymous stateid, which can be used in READ, WRITE, and SETATTR requests to indicate the absence of any open state associated with the request. When an anonymous stateid value is used, and an existing open denies the form of access requested, then access will be denied to the request. o When "other" and "seqid" are both all ones, the stateid is a special read bypass stateid. When this value is used in WRITE or SETATTR, it is treated like the anonymous value. When used in READ, the server MAY grant access, even if access would normally be denied to READ requests. o When "other" is zero and "seqid" is one, the stateid represents the current stateid, which is whatever value is the last stateid returned by an operation within the COMPOUND. In the case of an OPEN, the stateid returned for the open file, and not the delegation is used. The stateid passed to the operation in place of the special value has its "seqid" value set to zero. If there is no operation in the COMPOUND which has returned a stateid value, the server MUST return the error NFS4ERR_BAD_STATEID. If a stateid value is used which has all zero or all ones in the "other" field, but does not match one of the cases above, the server MUST return the error NFS4ERR_BAD_STATEID. Special stateids, unlike other stateids are not associated with individual client ID's or filehandles and can be used with all valid Shepler, et al. Expires September 5, 2007 [Page 132] Internet-Draft NFSv4 Minor Version 1 March 2007 client ID's and filehandles. In the case of a special stateid designating the current current stateid, the current stateid value substituted for the special stateid is associated with a particular client ID and filehandle. 8.1.3.3. Stateid Lifetime and Validation Stateids must remain valid until either a client reboot or a sever reboot or until the client returns all of the locks associated with the stateid by means of an operation such as CLOSE or DELEGRETURN. If the locks are lost due to revocation the stateid remains a valid designation of that revoked state until the client frees it by using FREE_STATEID. Stateids associated with record locks are an exception. They remain valid even if a LOCKU free all remaining locks, so long as the open file with which they are associated remains open, unless the client does a FREE_STATEID to cause the stateid to be freed. An "other" value must never be reused for a different purpose (i.e. different filehandle, owner, or type of locks) within the context of a single client ID. A server may retain the "other" value for the same purpose beyond the point where it may otherwise be freed but if it does so, it must maintain "seqid" continuity with previous values, in all case in which it is required to return incrementing "seqid" values in general. One mechanism that may be used to satisfy the requirement that the server recognize invalid and out-of-date stateids is for the server to divide the "other" field of the stateid into two fields. o An index into a table of locking-state structures. o A generation number which is incremented on each allocation of a table entry for a particular use. And then store in each table entry, o The current generation number. o The client ID with which the stateid is associated. o The filehandle of the file on which the locks are taken. o An indication of the type of stateid (open, record lock, file delegation, directory delegation, layout). o The last "seqid" value returned corresponding to the current "other" value. Shepler, et al. Expires September 5, 2007 [Page 133] Internet-Draft NFSv4 Minor Version 1 March 2007 With this information, the following procedure would be used to validate an incoming stateid and return an appropriate error, when necessary: o If the server has restarted resulting in loss of all lessed state but the sessionid and clientID are still valid, return NFS4ERR_STALE_STATEID. (If server restart has resulted in an invalid client ID or sessionid is invalid, SEQUENCE will return an error - not NFS4ERR_STATE_STATEID - and the operation that takes a stateid as an argument will never be processed.) o If the "other" field is all zeros or all ones, check that the "other" and "seqid" match a defined combination for a special stateid and that that stateid can be used in the current context. If not, then return NFS4ERR_BAD_STATEID. o If the "seqid" field is not zero, return NFS4ERR_BAD_STATEID. o Otherwise divide the "other" into a table index and an entry generation. o If the table index field is outside the range of the associated table, return NFS4ERR_BAD_STATEID. o If the selected table entry is of a different generation than that specified in the incoming stateid, return NFS4ERR_BAD_STATEID. o If the selected table entry does not match the current file handle, return NFS4ERR_BAD_STATEID. o If the client ID in the table entry does not match the client ID associated with the current session, return NFS4ERR_BAD_STATEID. o If the stateid type is not valid for the context in which the stateid appears, return NFS4ERR_BAD_STATEID. o Otherwise, the stateid is valid and the table entry should contain any additional information about the associated set of locks, such as open-owner and lock-owner information, as well as information on the specific locks, such as open modes and octet ranges. 8.1.4. Use of the Stateid and Locking All READ, WRITE and SETATTR operations contain a stateid. For the purposes of this section, SETATTR operations which change the size attribute of a file are treated as if they are writing the area between the old and new size (i.e. the range truncated or added to the file by means of the SETATTR), even where SETATTR is not Shepler, et al. Expires September 5, 2007 [Page 134] Internet-Draft NFSv4 Minor Version 1 March 2007 explicitly mentioned in the text. If the state-owner performs a READ or WRITE in a situation in which it has established a lock or share reservation on the server (any OPEN constitutes a share reservation) the stateid (previously returned by the server) must be used to indicate what locks, including both record locks and share reservations, are held by the state-owner. If no state is established by the client, either record lock or share reservation, a special stateid for anonymous state (zero as "other" and "seqid") is used. Regardless whether a stateid for anonymous state or a stateid returned by the server is used, if there is a conflicting share reservation or mandatory record lock held on the file, the server MUST refuse to service the READ or WRITE operation. Share reservations are established by OPEN operations and by their nature are mandatory in that when the OPEN denies READ or WRITE operations, that denial results in such operations being rejected with error NFS4ERR_LOCKED. Record locks may be implemented by the server as either mandatory or advisory, or the choice of mandatory or advisory behavior may be determined by the server on the basis of the file being accessed (for example, some UNIX-based servers support a "mandatory lock bit" on the mode attribute such that if set, record locks are required on the file before I/O is possible). When record locks are advisory, they only prevent the granting of conflicting lock requests and have no effect on READs or WRITEs. Mandatory record locks, however, prevent conflicting I/O operations. When they are attempted, they are rejected with NFS4ERR_LOCKED. When the client gets NFS4ERR_LOCKED on a file it knows it has the proper share reservation for, it will need to issue a LOCK request on the region of the file that includes the region the I/O was to be performed on, with an appropriate locktype (i.e. READ*_LT for a READ operation, WRITE*_LT for a WRITE operation). Note that for UNIX environments that support mandatory file locking, the distinction between advisory and mandatory locking is subtle. In fact, advisory and mandatory record locks are exactly the same in so far as the APIs and requirements on implementation. If the mandatory lock attribute is set on the file, the server checks to see if the lock-owner has an appropriate shared (read) or exclusive (write) record lock on the region it wishes to read or write to. If there is no appropriate lock, the server checks if there is a conflicting lock (which can be done by attempting to acquire the conflicting lock on the behalf of the lock-owner, and if successful, release the lock after the READ or WRITE is done), and if there is, the server returns NFS4ERR_LOCKED. For Windows environments, there are no advisory record locks, so the Shepler, et al. Expires September 5, 2007 [Page 135] Internet-Draft NFSv4 Minor Version 1 March 2007 server always checks for record locks during I/O requests. Thus, the NFS version 4 LOCK operation does not need to distinguish between advisory and mandatory record locks. It is the NFS version 4 server's processing of the READ and WRITE operations that introduces the distinction. Every stateid with the exception of special stateid values, whether returned by an OPEN-type operation (i.e. OPEN, OPEN_DOWNGRADE), or by a LOCK-type operation (i.e. LOCK or LOCKU), defines an access mode for the file (i.e. READ, WRITE, or READ-WRITE) as established by the original OPEN which caused the allocation of the open stateid and as modified by subsequent OPENs and OPEN_DOWNGRADEs for the same open-owner/file pair. Stateids returned by record lock operations imply the access mode for the open stateid associated with the lock set represented by the stateid. Delegation stateids have an access mode based on the type of delegation. When a READ, WRITE, or SETATTR which specifies the size attribute, is done, the operation is subject to checking against the access mode to verify that the operation is appropriate given the OPEN with which the operation is associated. In the case of WRITE-type operations (i.e. WRITEs and SETATTRs which set size), the server must verify that the access mode allows writing and return an NFS4ERR_OPENMODE error if it does not. In the case, of READ, the server may perform the corresponding check on the access mode, or it may choose to allow READ on opens for WRITE only, to accommodate clients whose write implementation may unavoidably do reads (e.g. due to buffer cache constraints). However, even if READs are allowed in these circumstances, the server MUST still check for locks that conflict with the READ (e.g. another open specify denial of READs). Note that a server which does enforce the access mode check on READs need not explicitly check for conflicting share reservations since the existence of OPEN for read access guarantees that no conflicting share reservation can exist. The read bypass special stateid (all bits of "other" and "seqid" set to one) stateid indicates a desire to bypass locking checks. The server MAY allow READ operations to bypass locking checks at the server, when this special stateid is used. However, WRITE operations with this special stateid value MUST NOT bypass locking checks and are treated exactly the same as if a special stateid for anonymous state were used. A lock may not be granted while a READ or WRITE operation using one of the special stateids is being performed and the range of the lock request conflicts with the range of the READ or WRITE operation. For the purposes of this paragraph, a conflict occurs when a shared lock is requested and a WRITE operation is being performed, or an Shepler, et al. Expires September 5, 2007 [Page 136] Internet-Draft NFSv4 Minor Version 1 March 2007 exclusive lock is requested and either a READ or a WRITE operation is being performed. A SETATTR that sets size is treated similarly to a WRITE as discussed above. 8.2. Lock Ranges The protocol allows a lock owner to request a lock with an octet range and then either upgrade, downgrade, or unlock a sub-range of the initial lock. It is expected that this will be an uncommon type of request. In any case, servers or server filesystems may not be able to support sub-range lock semantics. In the event that a server receives a locking request that represents a sub-range of current locking state for the lock owner, the server is allowed to return the error NFS4ERR_LOCK_RANGE to signify that it does not support sub- range lock operations. Therefore, the client should be prepared to receive this error and, if appropriate, report the error to the requesting application. The client is discouraged from combining multiple independent locking ranges that happen to be adjacent into a single request since the server may not support sub-range requests and for reasons related to the recovery of file locking state in the event of server failure. As discussed in the section "Server Failure and Recovery" below, the server may employ certain optimizations during recovery that work effectively only when the client's behavior during lock recovery is similar to the client's locking behavior prior to server failure. 8.3. Upgrading and Downgrading Locks If a client has a write lock on a record, it can request an atomic downgrade of the lock to a read lock via the LOCK request, by setting the type to READ_LT. If the server supports atomic downgrade, the request will succeed. If not, it will return NFS4ERR_LOCK_NOTSUPP. The client should be prepared to receive this error, and if appropriate, report the error to the requesting application. If a client has a read lock on a record, it can request an atomic upgrade of the lock to a write lock via the LOCK request by setting the type to WRITE_LT or WRITEW_LT. If the server does not support atomic upgrade, it will return NFS4ERR_LOCK_NOTSUPP. If the upgrade can be achieved without an existing conflict, the request will succeed. Otherwise, the server will return either NFS4ERR_DENIED or NFS4ERR_DEADLOCK. The error NFS4ERR_DEADLOCK is returned if the client issued the LOCK request with the type set to WRITEW_LT and the server has detected a deadlock. The client should be prepared to receive such errors and if appropriate, report the error to the requesting application. Shepler, et al. Expires September 5, 2007 [Page 137] Internet-Draft NFSv4 Minor Version 1 March 2007 8.4. Blocking Locks Some clients require the support of blocking locks. NFSv4.1 does not provide a callback when a previously unavailable lock becomes available. Clients thus have no choice but to continually poll for the lock. This presents a fairness problem. Two new lock types are added, READW and WRITEW, and are used to indicate to the server that the client is requesting a blocking lock. The server should maintain an ordered list of pending blocking locks. When the conflicting lock is released, the server may wait the lease period for the first waiting client to re-request the lock. After the lease period expires the next waiting client request is allowed the lock. Clients are required to poll at an interval sufficiently small that it is likely to acquire the lock in a timely manner. The server is not required to maintain a list of pending blocked locks as it is used to increase fairness and not correct operation. Because of the unordered nature of crash recovery, storing of lock state to stable storage would be required to guarantee ordered granting of blocking locks. Servers may also note the lock types and delay returning denial of the request to allow extra time for a conflicting lock to be released, allowing a successful return. In this way, clients can avoid the burden of needlessly frequent polling for blocking locks. The server should take care in the length of delay in the event the client retransmits the request. If a server receives a blocking lock request, denies it, and then later receives a nonblocking request for the same lock, which is also denied, then it should remove the lock in question from its list of pending blocking locks. Clients should use such a nonblocking request to indicate to the server that this is the last time they intend to poll for the lock, as may happen when the process requesting the lock is interrupted. This is a courtesy to the server, to prevent it from unnecessarily waiting a lease period before granting other lock requests. However, clients are not required to perform this courtesy, and servers must not depend on them doing so. Also, clients must be prepared for the possibility that this final locking request will be accepted. 8.5. Lease Renewal The purpose of a lease is to allow a server to remove stale locks that are held by a client that has crashed or is otherwise unreachable. It is not a mechanism for cache consistency and lease renewals may not be denied if the lease interval has not expired. Since each session is associated with a specific client, any Shepler, et al. Expires September 5, 2007 [Page 138] Internet-Draft NFSv4 Minor Version 1 March 2007 operation issued on that session is an indication that the associated client is reachable. When a request is issued for a given session, execution of a SEQUENCE operation will result in all leases for the associated client to be implicitly renewed. This approach allows for low overhead lease renewal which scales well. In the typical case no extra RPC calls are required for lease renewal and in the worst case one RPC is required every lease period, via a COMPOUND that consists solely of a single SEQUENCE operation. The number of locks held by the client is not a factor since all state for the client is involved with the lease renewal action. Since all operations that create a new lease also renew existing leases, the server must maintain a common lease expiration time for all valid leases for a given client. This lease time can then be easily updated upon implicit lease renewal actions. 8.6. Crash Recovery The important requirement in crash recovery is that both the client and the server know when the other has failed. Additionally, it is required that a client sees a consistent view of data across server restarts or reboots. All READ and WRITE operations that may have been queued within the client or network buffers must wait until the client has successfully recovered the locks protecting the READ and WRITE operations. 8.6.1. Client Failure and Recovery In the event that a client fails, the server may release the client's locks when the associated leases have expired. Conflicting locks from another client may only be granted after this lease expiration. When a client has not failed and re-establishes his lease before expiration occurs, requests for conflicting locks will not be granted. To minimize client delay upon restart, lock requests are associated with an instance of the client by a client supplied verifier. This verifier is part of the initial EXCHANGE_ID call made by the client. The server returns a client ID as a result of the EXCHANGE_ID operation. The client then confirms the use of the client ID by establishing a session associated with that client ID. All locks, including opens, record locks, delegations, and layout obtained by sessions using that client ID are associated with that client ID. Since the verifier will be changed by the client upon each initialization, the server can compare a new verifier to the verifier associated with currently held locks and determine that they do not match. This signifies the client's new instantiation and subsequent Shepler, et al. Expires September 5, 2007 [Page 139] Internet-Draft NFSv4 Minor Version 1 March 2007 loss of locking state. As a result, the server is free to release all locks held which are associated with the old client ID which was derived from the old verifier. At this point conflicting locks from other clients, kept waiting while the leaser had not yet expired, can be granted. Note that the verifier must have the same uniqueness properties of the verifier for the COMMIT operation. 8.6.2. Server Failure and Recovery If the server loses locking state (usually as a result of a restart or reboot), it must allow clients time to discover this fact and re- establish the lost locking state. The client must be able to re- establish the locking state without having the server deny valid requests because the server has granted conflicting access to another client. Likewise, if there is a possibility that clients have not yet re-established their locking state for a file, the server must disallow READ and WRITE operations for that file. A client can determine that loss of locking state has occurred via several methods. 1. When a SEQUENCE succeeds, but sr_status_flags in the reply to SEQUENCE indicates SEQ4_STATUS_RESTART_RECLAIM_NEEDED (see Section 17.46.4). The client's client ID and session are valid (have persisted through server restart) and the client can now re-establish its lock state (Section 8.6.2.1). 2. When an operation returns NFS4ERR_STALE_STATEID, this indicates a stateid invalidated by a server reboot or restart. Since the operation that returned NFS4ERR_STALE_STATEID MUST have been preceded by SEQUENCE, and SEQUENCE did not return an error, this means the client ID and session are valid. The client can now re-establish is lock state as described in Section 8.6.2.1. Note that the server should (MUST) have set SEQ4_STATUS_RESTART_RECLAIM_NEEDED in the sr_status_flags of the results of the SEQUENCE operation, and thus this situation should be the same as that described above. 3. When a SEQUENCE operation returns NFS4ERR_STALE_CLIENTID, this means both sessionid SEQUENCE refers to (field sa_sessionid) and the implied client ID are now invalid, where the client ID was invalidated by server reboot or restart or by lease expiration. When SEQUENCE returns NFS4ERR_STALE_CLIENTID, the client must establish a new client ID (see Section 8.1.1) and re-establish its lock state (Section 8.6.2.1). Shepler, et al. Expires September 5, 2007 [Page 140] Internet-Draft NFSv4 Minor Version 1 March 2007 4. When a SEQUENCE operation returns NFS4ERR_BADSESSION, this may mean the session has been destroyed, but the client ID is still valid. The client issues a CREATE_SESSION request with the client ID to re-establish the session. If CREATE_SESSION fails with NFS4ERR_STALE_CLIENTID, the client must establish a new client ID (see Section 8.1.1) and re-establish its lock state (Section 8.6.2.1). If CREATE_SESSION succeeds, the client must then re-establish its lock state (Section 8.6.2.1). 5. When a operation, neither SEQUENCE nor preceded by SEQUENCE (for example, CREATE_SESSION, DESTROY_SESSION) returns NFS4ERR_STALE_CLIENTID. The client MUST establish a new client ID (Section 8.1.1) and re-establish its lock state (Section 8.6.2.1). 8.6.2.1. State Reclaim Once a session is established using the new client ID, the client will use reclaim-type locking requests (i.e. LOCK requests with reclaim set to true and OPEN operations with a claim type of CLAIM_PREVIOUS) to re-establish its locking state. Once this is done, or if there is no such locking state to reclaim, the client does a RECLAIM_COMPLETE operation to indicate that it has reclaimed all of the locking state that it will reclaim. Once a client does a RECLAIM_COMPLETE operation, it may attempt non-reclaim locking operations, although it may get NFS4ERR_GRACE errors on these until the period of special handling is over. The period of special handling of locking and READs and WRITEs, is referred to as the "grace period". During the grace period, clients recover locks and the associated state using reclaim-type locking requests. During this period, the server must reject READ and WRITE operations and non-reclaim locking requests (i.e. other LOCK and OPEN operations) with an error of NFS4ERR_GRACE, unless it is able to guarantee that these may be done safely, as described below. The grace period may last until all clients who are known to possibly have had locks have done a RECLAIM_COMPLETE operation, indicating that they have finished reclaiming the locks they held before the server reboot. The server is assumed to maintain in stable storage a list of clients who may have such locks. The server may also terminate the grace period before all clients have done RECLAIM_COMPLETE. The server SHOULD NOT terminate the grace period before a time equal to the lease period in order to give clients an opportunity to find out about the server reboot. Some additional time in order to allow time to establish a new client ID and session and to effect lock reclaims may be added. Shepler, et al. Expires September 5, 2007 [Page 141] Internet-Draft NFSv4 Minor Version 1 March 2007 If the server can reliably determine that granting a non-reclaim request will not conflict with reclamation of locks by other clients, the NFS4ERR_GRACE error does not have to be returned even within the grace period, although NFS4ERR_GRACE must always be returned to clients attempting a non-reclaim lock request before doing their own RECLAIM_COMPLETE. For the server to be able to service READ and WRITE operations during the grace period, it must again be able to guarantee that no possible conflict could arise between a potential reclaim locking request and the READ or WRITE operation. If the server is unable to offer that guarantee, the NFS4ERR_GRACE error must be returned to the client. For a server to provide simple, valid handling during the grace period, the easiest method is to simply reject all non-reclaim locking requests and READ and WRITE operations by returning the NFS4ERR_GRACE error. However, a server may keep information about granted locks in stable storage. With this information, the server could determine if a regular lock or READ or WRITE operation can be safely processed. For example, if the server maintained on stable storage summary information on whether mandatory locks exist, either mandatory record locks, or share reservations specifying deny modes, many requests could be allowed during the grace period. If it is known that no such share reservations exist, OPEN request that do not specify deny modes may be safely granted. If, in addition, it is known that no mandatory record locks exist, either through information stored on stable storage or simply because the server does not support such locks, READ and WRITE requests may be safely processed during the grace period. To reiterate, for a server that allows non-reclaim lock and I/O requests to be processed during the grace period, it MUST determine that no lock subsequently reclaimed will be rejected and that no lock subsequently reclaimed would have prevented any I/O operation processed during the grace period. Clients should be prepared for the return of NFS4ERR_GRACE errors for non-reclaim lock and I/O requests. In this case the client should employ a retry mechanism for the request. A delay (on the order of several seconds) between retries should be used to avoid overwhelming the server. Further discussion of the general issue is included in [Floyd]. The client must account for the server that is able to perform I/O and non-reclaim locking requests within the grace period as well as those that can not do so. A reclaim-type locking request outside the server's grace period can only succeed if the server can guarantee that no conflicting lock or Shepler, et al. Expires September 5, 2007 [Page 142] Internet-Draft NFSv4 Minor Version 1 March 2007 I/O request has been granted since reboot or restart. A server may, upon restart, establish a new value for the lease period. Therefore, clients should, once a new client ID is established, refetch the lease_time attribute and use it as the basis for lease renewal for the lease associated with that server. However, the server must establish, for this restart event, a grace period at least as long as the lease period for the previous server instantiation. This allows the client state obtained during the previous server instance to be reliably re-established. 8.6.3. Network Partitions and Recovery If the duration of a network partition is greater than the lease period provided by the server, the server will have not received a lease renewal from the client. If this occurs, the server may free all locks held for the client, or it may allow the lock state to remain for a considerable period, subject to the constraint that if a request for a conflicting lock is made, locks associated with expired leases do not prevent such a conflicting lock from being granted but are revoked as necessary so as not to interfere with such conflicting requests. If the server chooses to delay freeing of lock state until there is a conflict, it may either free all of the clients locks once there is a conflict, or it may only revoke the minimum set of locks necessary to allow conflicting requests. When it adopts the finer-grained approach, it must revoke all locks associated with a given stateid, as long as it revokes a single such lock. When the server chooses to free all of a client's lock state, either immediately upon lease expiration, or a result of the first attempt to get a lock, all stateids held by the client will become invalid or stale. Once the client is able to reach the server after such a network partition, the status returned by the SEQUENCE operation will indicate a loss of locking state. In addition all I/O submitted by the client with the now invalid stateids will fail with the server returning the error NFS4ERR_EXPIRED. Once the client learns of the loss of locking state, it will suitably notify the applications that held the invalidated locks. The client should then take action to free invalidated stateids, either by establishing a new client ID using a new verifier or by doing a FREE_STATEID operation to release each of the invalidated stateids. When the server adopts a finer-grained approach to revocation of locks when lease have expired, only a subset of stateids will normally become invalid during a network partition. When the client is able to communicate with the server after such a network Shepler, et al. Expires September 5, 2007 [Page 143] Internet-Draft NFSv4 Minor Version 1 March 2007 partition, the status returned by the SEQUENCE operation will indicate a partial loss of locking state. In addition, operations, including I/O submitted by the client with the now invalid stateids will fail with the server returning the error NFS4ERR_EXPIRED. Once the client learns of the loss of locking state, it will use the TEST_STATEID operation on all of its stateids to determine which locks have been lost and them suitably notify the applications that held the invalidated locks. The client can then release the invalidated locking state and acknowledge the revocation of the associated locks by doing a FREE_STATEID operation on each of the invalidated stateids. When a network partition is combined with a server reboot, there are edge conditions that place requirements on the server in order to avoid silent data corruption following the server reboot. Two of these edge conditions are known, and are discussed below. The first edge condition arises as a result of the scenarios such as the following: 1. Client A acquires a lock. 2. Client A and server experience mutual network partition, such that client A is unable to renew its lease. 3. Client A's lease expires, and the server releases lock. 4. Client B acquires a lock that would have conflicted with that of Client A. 5. Client B releases its lock. 6. Server reboots. 7. Network partition between client A and server heals. 8. Client A connects to new server instance and finds out about server reboot. 9. Client A reclaims its lock within the server's grace period. Thus, at the final step, the server has erroneously granted client A's lock reclaim. If client B modified the object the lock was protecting, client A will experience object corruption. The second known edge condition arises in situations such as the following: Shepler, et al. Expires September 5, 2007 [Page 144] Internet-Draft NFSv4 Minor Version 1 March 2007 1. Client A acquires one or more locks. 2. Server reboots. 3. Client A and server experience mutual network partition, such that client A is unable to reclaim all of its locks within the grace period. 4. Server's reclaim grace period ends. Client A has either no locks or an incomplete set of locks known to the server. 5. Client B acquires a lock that would have conflicted with a lock of client A that was not reclaimed. 6. Client B releases the lock. 7. Server reboots a second time. 8. Network partition between client A and server heals. 9. Client A connects to new server instance and finds out about server reboot. 10. Client A reclaims its lock within the server's grace period. As with the first edge condition, the final step of the scenario of the second edge condition has the server erroneously granting client A's lock reclaim. Solving the first and second edge conditions requires that the server either always assumes after it reboots that some edge condition occurs, and thus return NFS4ERR_NO_GRACE for all reclaim attempts, or that the server record some information in stable storage. The amount of information the server records in stable storage is in inverse proportion to how harsh the server intends to be whenever edge conditions arise. The server that is completely tolerant of all edge conditions will record in stable storage every lock that is acquired, removing the lock record from stable storage only when the lock is released. For the two edge conditions discussed above, the harshest a server can be, and still support a grace period for reclaims, requires that the server record in stable storage information some minimal information. For example, a server implementation could, for each client, save in stable storage a record containing: o the client's id string Shepler, et al. Expires September 5, 2007 [Page 145] Internet-Draft NFSv4 Minor Version 1 March 2007 o a boolean that indicates if the client's lease expired or if there was administrative intervention (see Section 8.7) to revoke a record lock, share reservation, or delegation and there has been no acknowledgement (via FREE_STATEID) of such revocation. o a boolean that indicates whether the client may have locks that it believes to be reclaimable in situations which the grace period was terminated, making the server's view of lock reclaimability suspect. The server will set this for any client record in stable storage where the client has not done a RECLAIM_COMPLETE, before it grants any new (i.e. not reclaimed) lock to any client. Assuming the above record keeping, for the first edge condition, after the server reboots, the record that client A's lease expired means that another client could have acquired a conflicting record lock, share reservation, or delegation. Hence the server must reject a reclaim from client A with the error NFS4ERR_NO_GRACE. For the second edge condition, after the server reboots for a second time, the indication that the client had not completed its reclaims at the time at which the grace period ended means that the server must reject a reclaim from client A with the error NFS4ERR_NO_GRACE. When either edge condition occurs, the client's attempt to reclaim locks will result in the error NFS4ERR_NO_GRACE. When this is received, or after the client reboots with no lock state, the client will issue a RECLAIM_COMPLETE. When the RECLAIM_COMPLETE is received, the server and client are again in agreement regarding reclaimable locks and both booleans in persistent storage can be reset, to be set again only when there is a subsequent event that causes lock reclaim operations to be questionable. Regardless of the level and approach to record keeping, the server MUST implement one of the following strategies (which apply to reclaims of share reservations, record locks, and delegations): 1. Reject all reclaims with NFS4ERR_NO_GRACE. This is extremely unforgiving, but necessary if the server does not record lock state in stable storage. 2. Record sufficient state in stable storage such that all known edge conditions involving server reboot, including the two noted in this section, are detected. False positives are acceptable. Note that at this time, it is not known if there are other edge conditions. In the event that, after a server reboot, the server determines that there is unrecoverable damage or corruption to the Shepler, et al. Expires September 5, 2007 [Page 146] Internet-Draft NFSv4 Minor Version 1 March 2007 information in stable storage, then for all clients and/or locks which may be affected, the server MUST return NFS4ERR_NO_GRACE. A mandate for the client's handling of the NFS4ERR_NO_GRACE error is outside the scope of this specification, since the strategies for such handling are very dependent on the client's operating environment. However, one potential approach is described below. When the client receives NFS4ERR_NO_GRACE, it could examine the change attribute of the objects the client is trying to reclaim state for, and use that to determine whether to re-establish the state via normal OPEN or LOCK requests. This is acceptable provided the client's operating environment allows it. In other words, the client implementor is advised to document for his users the behavior. The client could also inform the application that its record lock or share reservations (whether they were delegated or not) have been lost, such as via a UNIX signal, a GUI pop-up window, etc. See the section, "Data Caching and Revocation" for a discussion of what the client should do for dealing with unreclaimed delegations on client state. For further discussion of revocation of locks see Section 8.7. 8.7. Server Revocation of Locks At any point, the server can revoke locks held by a client and the client must be prepared for this event. When the client detects that its locks have been or may have been revoked, the client is responsible for validating the state information between itself and the server. Validating locking state for the client means that it must verify or reclaim state for each lock currently held. The first occasion of lock revocation is upon server reboot or restart. In this instance the client will receive an error (NFS4ERR_STALE_STATEID on an operation that takes a stateid as an argument or NFS4ERR_STALE_CLIENTID on an operation that takes a sessionid or client ID) and the client will proceed with normal crash recovery as described in the Section 8.6.2.1. The second occasion of lock revocation is the inability to renew the lease before expiration, as discussed above. While this is considered a rare or unusual event, the client must be prepared to recover. The server is responsible for determining lease expiration, and deciding exactly how to deal with it, informing the client of the scope of the lock revocation. The client then uses the status information provided by the server in the SEQUENCE results (field sr_status_flags, see Section 17.46.4) to synchronize its locking state with that of the server, in order to recover. Shepler, et al. Expires September 5, 2007 [Page 147] Internet-Draft NFSv4 Minor Version 1 March 2007 The third occasion of lock revocation can occur as a result of revocation of locks within the lease period, either because of administrative intervention, or because a recallable lock (a delegation or layout) was not returned within the lease period after having been recalled. While these are considered rare events, they are possible and the client must be prepared to deal with them. When either of these events occur, the client finds out about the situation through the status returned by the SEQUENCE operation. Any use of stateids associated with revoked locks will receive the error NFS4ERR_ADMIN_REVOKED or NFS4ERR_DELEG_REVOKED, as appropriate. In all situations in which a subset of locking state may have been revoked, which include all cases in which locking state is revoked within the lease period, it is up to the client to determine which locks have been revoked and which have not. It does this by using the TEST_STATEID operation on the appropriate set of stateids. Once the set of revoked locks has been determined, the applications can be notified, and the invalidated stateids can be freed and lock revocation acknowledged by using FREE_STATEID. 8.8. Share Reservations A share reservation is a mechanism to control access to a file. It is a separate and independent mechanism from record locking. When a client opens a file, it issues an OPEN operation to the server specifying the type of access required (READ, WRITE, or BOTH) and the type of access to deny others (deny NONE, READ, WRITE, or BOTH). If the OPEN fails the client will fail the application's open request. Pseudo-code definition of the semantics: if (request.access == 0) return (NFS4ERR_INVAL) else if ((request.access & file_state.deny)) || (request.deny & file_state.access)) return (NFS4ERR_DENIED) This checking of share reservations on OPEN is done with no exception for an existing OPEN for the same open-owner. The constants used for the OPEN and OPEN_DOWNGRADE operations for the access and deny fields are as follows: Shepler, et al. Expires September 5, 2007 [Page 148] Internet-Draft NFSv4 Minor Version 1 March 2007 const OPEN4_SHARE_ACCESS_READ = 0x00000001; const OPEN4_SHARE_ACCESS_WRITE = 0x00000002; const OPEN4_SHARE_ACCESS_BOTH = 0x00000003; const OPEN4_SHARE_DENY_NONE = 0x00000000; const OPEN4_SHARE_DENY_READ = 0x00000001; const OPEN4_SHARE_DENY_WRITE = 0x00000002; const OPEN4_SHARE_DENY_BOTH = 0x00000003; 8.9. OPEN/CLOSE Operations To provide correct share semantics, a client MUST use the OPEN operation to obtain the initial filehandle and indicate the desired access and what access, if any, to deny. Even if the client intends to use a special stateid for anonymous state or read bypass, it must still obtain the filehandle for the regular file with the OPEN operation so the appropriate share semantics can be applied. For clients that do not have a deny mode built into their open programming interfaces, deny equal to NONE should be used. The OPEN operation with the CREATE flag, also subsumes the CREATE operation for regular files as used in previous versions of the NFS protocol. This allows a create with a share to be done atomically. The CLOSE operation removes all share reservations held by the open- owner on that file. If record locks are held, the client SHOULD release all locks before issuing a CLOSE. The server MAY free all outstanding locks on CLOSE but some servers may not support the CLOSE of a file that still has record locks held. The server MUST return failure, NFS4ERR_LOCKS_HELD, if any locks would exist after the CLOSE. The LOOKUP operation will return a filehandle without establishing any lock state on the server. Without a valid stateid, the server will assume the client has the least access. For example, a file opened with deny READ/WRITE using a filehandle obtained through LOOKUP could only be read using the special read bypass stateid and could not be written at all because it would not have a valid stateid and the special anonymous stateid would not be allowed access. 8.10. Open Upgrade and Downgrade When an OPEN is done for a file and the open-owner for which the open is being done already has the file open, the result is to upgrade the open file status maintained on the server to include the access and deny bits specified by the new OPEN as well as those for the existing OPEN. The result is that there is one open file, as far as the protocol is concerned, and it includes the union of the access and Shepler, et al. Expires September 5, 2007 [Page 149] Internet-Draft NFSv4 Minor Version 1 March 2007 deny bits for all of the OPEN requests completed. Only a single CLOSE will be done to reset the effects of both OPENs. Note that the client, when issuing the OPEN, may not know that the same file is in fact being opened. The above only applies if both OPENs result in the OPENed object being designated by the same filehandle. When the server chooses to export multiple filehandles corresponding to the same file object and returns different filehandles on two different OPENs of the same file object, the server MUST NOT "OR" together the access and deny bits and coalesce the two open files. Instead the server must maintain separate OPENs with separate stateids and will require separate CLOSEs to free them. When multiple open files on the client are merged into a single open file object on the server, the close of one of the open files (on the client) may necessitate change of the access and deny status of the open file on the server. This is because the union of the access and deny bits for the remaining opens may be smaller (i.e. a proper subset) than previously. The OPEN_DOWNGRADE operation is used to make the necessary change and the client should use it to update the server so that share reservation requests by other clients are handled properly. 8.11. Short and Long Leases When determining the time period for the server lease, the usual lease tradeoffs apply. Short leases are good for fast server recovery at a cost of increased operations to effect lease renewal (when there are no other operations during the period to effect lease renewal as a side-effect). Long leases are certainly kinder and gentler to servers trying to handle very large numbers of clients. The number of extra requests to effect lock renewal drop in inverse proportion to the lease time. The disadvantages of long leases include the possibility of slower recovery after certain failures. After server failure, a longer grace period may be required when some clients do not promptly reclaim their locks and do a RECLAIM_COMPLETE. In the event of client failure, it can longer period for leases to expire thus forcing conflicting requests to wait. Long leases are usable if the server is able to store lease state in non-volatile memory. Upon recovery, the server can reconstruct the lease state from its non-volatile memory and continue operation with its clients and therefore long leases would not be an issue. Shepler, et al. Expires September 5, 2007 [Page 150] Internet-Draft NFSv4 Minor Version 1 March 2007 8.12. Clocks, Propagation Delay, and Calculating Lease Expiration To avoid the need for synchronized clocks, lease times are granted by the server as a time delta. However, there is a requirement that the client and server clocks do not drift excessively over the duration of the lock. There is also the issue of propagation delay across the network which could easily be several hundred milliseconds as well as the possibility that requests will be lost and need to be retransmitted. To take propagation delay into account, the client should subtract it from lease times (e.g. if the client estimates the one-way propagation delay as 200 msec, then it can assume that the lease is already 200 msec old when it gets it). In addition, it will take another 200 msec to get a response back to the server. So the client must send a lock renewal or write data back to the server 400 msec before the lease would expire. The server's lease period configuration should take into account the network distance of the clients that will be accessing the server's resources. It is expected that the lease period will take into account the network propagation delays and other network delay factors for the client population. Since the protocol does not allow for an automatic method to determine an appropriate lease period, the server's administrator may have to tune the lease period. 8.13. Vestigial Locking Infrastructure From V4.0 There are a number of operations and fields within existing operations that no longer have a function in minor version one. In one way or another, these changes are all due to the implementation of sessions which provides client context and replay protection as a base feature of the protocol, separate from locking itself. The following operations have become mandatory-to-not-implement. The server should return NFS4ERR_NOTSUPP if these operations are found in an NFSv4.1 COMPOUND. o SETCLIENTID since its function has been replaced by EXCHANGE_ID. o SETCLIENTID_CONFIRM since client ID confirmation now happens by means of CREATE_SESSION. o OPEN_CONFIRM because OPEN's no longer require confirmation to establish an owner-based sequence value. o RELEASE_LOCKOWNER because lock-owners with no associated locks have any sequence-related state and so can be deleted by the Shepler, et al. Expires September 5, 2007 [Page 151] Internet-Draft NFSv4 Minor Version 1 March 2007 server at will. o RENEW because every SEQUENCE operation for a session causes lease renewal, making a separate operation useless. Also, there are a number of fields, present in existing operations related to locking that have no use in minor version one. They were used in minor version zero to perform functions now provided in a different fashion. o Sequence id's used to sequence requests for a given state-owner and to provide replay protection, now provided via sessions. o Client IDs used to identify the client associated with a given request. Client identification is now available using the client ID associated with the current session, without needing an explicit client ID field. Such vestigial fields in existing operations should be set by the client to zero. When they are not, the server MUST return an NFS4ERR_INVAL error. 9. Client-Side Caching Client-side caching of data, of file attributes, and of file names is essential to providing good performance with the NFS protocol. Providing distributed cache coherence is a difficult problem and previous versions of the NFS protocol have not attempted it. Instead, several NFS client implementation techniques have been used to reduce the problems that a lack of coherence poses for users. These techniques have not been clearly defined by earlier protocol specifications and it is often unclear what is valid or invalid client behavior. The NFS version 4 protocol uses many techniques similar to those that have been used in previous protocol versions. The NFS version 4 protocol does not provide distributed cache coherence. However, it defines a more limited set of caching guarantees to allow locks and share reservations to be used without destructive interference from client side caching. In addition, the NFS version 4 protocol introduces a delegation mechanism which allows many decisions normally made by the server to be made locally by clients. This mechanism provides efficient support of the common cases where sharing is infrequent or where sharing is read-only. Shepler, et al. Expires September 5, 2007 [Page 152] Internet-Draft NFSv4 Minor Version 1 March 2007 9.1. Performance Challenges for Client-Side Caching Caching techniques used in previous versions of the NFS protocol have been successful in providing good performance. However, several scalability challenges can arise when those techniques are used with very large numbers of clients. This is particularly true when clients are geographically distributed which classically increases the latency for cache revalidation requests. The previous versions of the NFS protocol repeat their file data cache validation requests at the time the file is opened. This behavior can have serious performance drawbacks. A common case is one in which a file is only accessed by a single client. Therefore, sharing is infrequent. In this case, repeated reference to the server to find that no conflicts exist is expensive. A better option with regards to performance is to allow a client that repeatedly opens a file to do so without reference to the server. This is done until potentially conflicting operations from another client actually occur. A similar situation arises in connection with file locking. Sending file lock and unlock requests to the server as well as the read and write requests necessary to make data caching consistent with the locking semantics (see the section "Data Caching and File Locking") can severely limit performance. When locking is used to provide protection against infrequent conflicts, a large penalty is incurred. This penalty may discourage the use of file locking by applications. The NFS version 4 protocol provides more aggressive caching strategies with the following design goals: .IP o Compatibility with a large range of server semantics. .IP o Provide the same caching benefits as previous versions of the NFS protocol when unable to provide the more aggressive model. .IP o Requirements for aggressive caching are organized so that a large portion of the benefit can be obtained even when not all of the requirements can be met. .LP The appropriate requirements for the server are discussed in later sections in which specific forms of caching are covered. (see the section "Open Delegation"). 9.2. Delegation and Callbacks Recallable delegation of server responsibilities for a file to a client improves performance by avoiding repeated requests to the server in the absence of inter-client conflict. With the use of a "callback" RPC from server to client, a server recalls delegated responsibilities when another client engages in sharing of a Shepler, et al. Expires September 5, 2007 [Page 153] Internet-Draft NFSv4 Minor Version 1 March 2007 delegated file. A delegation is passed from the server to the client, specifying the object of the delegation and the type of delegation. There are different types of delegations but each type contains a stateid to be used to represent the delegation when performing operations that depend on the delegation. This stateid is similar to those associated with locks and share reservations but differs in that the stateid for a delegation is associated with a client ID and may be used on behalf of all the open_owners for the given client. A delegation is made to the client as a whole and not to any specific process or thread of control within it. The callback path or backchannel is established by CREATE_SESSION and BIND_CONN_TO_SESSION, and the client is required to maintain it. Because the backchannel may be down, even temporarily, correct protocol operation does not depend on them. Preliminary testing of callback functionality by means of a CB_COMPOUND procedure with a single operation, CB_SEQUENCE, can be used to check the continuity of the backchannel. A server avoids delegating responsibilities until it has determined that the backchannel exists. Because the granting of a delegation is always conditional upon the absence of conflicting access, clients must not assume that a delegation will be granted and they must always be prepared for OPENs, WANT_DELEGATIONs, and GET_DIR_DELEGATIONs to be processed without any delegations being granted. Once granted, a delegation behaves in many ways like a lock. There is an associated lease that is subject to renewal together with all of the other leases held by that client. Unlike locks, an operation by a second client to a delegated file will cause the server to recall a delegation through a callback. On recall, the client holding the delegation must flush modified state (such as modified data) to the server and return the delegation. The conflicting request will not receive a response until the recall is complete. The recall is considered complete when the client returns the delegation or the server times out on the recall and revokes the delegation as a result of the timeout. Following the resolution of the recall, the server has the information necessary to grant or deny the second client's request. At the time the client receives a delegation recall, it may have substantial state that needs to be flushed to the server. Therefore, the server should allow sufficient time for the delegation to be returned since it may involve numerous RPCs to the server. If the server is able to determine that the client is diligently flushing Shepler, et al. Expires September 5, 2007 [Page 154] Internet-Draft NFSv4 Minor Version 1 March 2007 state to the server as a result of the recall, the server may extend the usual time allowed for a recall. However, the time allowed for recall completion should not be unbounded. An example of this is when responsibility to mediate opens on a given file is delegated to a client (see the section "Open Delegation"). The server will not know what opens are in effect on the client. Without this knowledge the server will be unable to determine if the access and deny state for the file allows any particular open until the delegation for the file has been returned. A client failure or a network partition can result in failure to respond to a recall callback. In this case, the server will revoke the delegation which in turn will render useless any modified state still on the client. 9.2.1. Delegation Recovery There are three situations that delegation recovery must deal with: o Client reboot or restart o Server reboot or restart o Network partition (full or callback-only) In the event the client reboots or restarts, the failure to renew leases will result in the revocation of record locks and share reservations. Delegations, however, may be treated a bit differently. There will be situations in which delegations will need to be reestablished after a client reboots or restarts. The reason for this is the client may have file data stored locally and this data was associated with the previously held delegations. The client will need to reestablish the appropriate file state on the server. To allow for this type of client recovery, the server MAY extend the period for delegation recovery beyond the typical lease expiration period. This implies that requests from other clients that conflict with these delegations will need to wait. Because the normal recall process may require significant time for the client to flush changed state to the server, other clients need be prepared for delays that occur because of a conflicting delegation. This longer interval would increase the window for clients to reboot and consult stable storage so that the delegations can be reclaimed. For open delegations, such delegations are reclaimed using OPEN with a claim type of CLAIM_DELEGATE_PREV. (See the sections on "Data Caching and Shepler, et al. Expires September 5, 2007 [Page 155] Internet-Draft NFSv4 Minor Version 1 March 2007 Revocation" and "Operation 18: OPEN" for discussion of open delegation and the details of OPEN respectively). A server MAY support a claim type of CLAIM_DELEGATE_PREV, but if it does, it MUST NOT remove delegations upon a CREATE_SESSION that confirms a client ID created by EXCHANGE_ID, and instead MUST, for a period of time no less than that of the value of the lease_time attribute, maintain the client's delegations to allow time for the client to issue CLAIM_DELEGATE_PREV requests. The server that supports CLAIM_DELEGATE_PREV MUST support the DELEGPURGE operation. When the server reboots or restarts, delegations are reclaimed (using the OPEN operation with CLAIM_PREVIOUS) in a similar fashion to record locks and share reservations. However, there is a slight semantic difference. In the normal case if the server decides that a delegation should not be granted, it performs the requested action (e.g. OPEN) without granting any delegation. For reclaim, the server grants the delegation but a special designation is applied so that the client treats the delegation as having been granted but recalled by the server. Because of this, the client has the duty to write all modified state to the server and then return the delegation. This process of handling delegation reclaim reconciles three principles of the NFS version 4 protocol: o Upon reclaim, a client reporting resources assigned to it by an earlier server instance must be granted those resources. o The server has unquestionable authority to determine whether delegations are to be granted and, once granted, whether they are to be continued. o The use of callbacks is not to be depended upon until the client has proven its ability to receive them. When a network partition occurs, delegations are subject to freeing by the server when the lease renewal period expires. This is similar to the behavior for locks and share reservations. For delegations, however, the server may extend the period in which conflicting requests are held off. Eventually the occurrence of a conflicting request from another client will cause revocation of the delegation. A loss of the callback path (e.g. by later network configuration change) will have the same effect. A recall request will fail and revocation of the delegation will result. A client normally finds out about revocation of a delegation when it uses a stateid associated with a delegation and receives the error NFS4ERR_EXPIRED. It also may find out about delegation revocation after a client reboot when it attempts to reclaim a delegation and Shepler, et al. Expires September 5, 2007 [Page 156] Internet-Draft NFSv4 Minor Version 1 March 2007 receives that same error. Note that in the case of a revoked write open delegation, there are issues because data may have been modified by the client whose delegation is revoked and separately by other clients. See the section "Revocation Recovery for Write Open Delegation" for a discussion of such issues. Note also that when delegations are revoked, information about the revoked delegation will be written by the server to stable storage (as described in the section "Crash Recovery"). This is done to deal with the case in which a server reboots after revoking a delegation but before the client holding the revoked delegation is notified about the revocation. 9.3. Data Caching When applications share access to a set of files, they need to be implemented so as to take account of the possibility of conflicting access by another application. This is true whether the applications in question execute on different clients or reside on the same client. Share reservations and record locks are the facilities the NFS version 4 protocol provides to allow applications to coordinate access by providing mutual exclusion facilities. The NFS version 4 protocol's data caching must be implemented such that it does not invalidate the assumptions that those using these facilities depend upon. 9.3.1. Data Caching and OPENs In order to avoid invalidating the sharing assumptions that applications rely on, NFS version 4 clients should not provide cached data to applications or modify it on behalf of an application when it would not be valid to obtain or modify that same data via a READ or WRITE operation. Furthermore, in the absence of open delegation (see the section "Open Delegation") two additional rules apply. Note that these rules are obeyed in practice by many NFS version 2 and version 3 clients. o First, cached data present on a client must be revalidated after doing an OPEN. Revalidating means that the client fetches the change attribute from the server, compares it with the cached change attribute, and if different, declares the cached data (as well as the cached attributes) as invalid. This is to ensure that the data for the OPENed file is still correctly reflected in the client's cache. This validation must be done at least when the client's OPEN operation includes DENY=WRITE or BOTH thus terminating a period in which other clients may have had the Shepler, et al. Expires September 5, 2007 [Page 157] Internet-Draft NFSv4 Minor Version 1 March 2007 opportunity to open the file with WRITE access. Clients may choose to do the revalidation more often (i.e. at OPENs specifying DENY=NONE) to parallel the NFS version 3 protocol's practice for the benefit of users assuming this degree of cache revalidation. Since the change attribute is updated for data and metadata modifications, some client implementors may be tempted to use the time_modify attribute and not change to validate cached data, so that metadata changes do not spuriously invalidate clean data. The implementor is cautioned in this approach. The change attribute is guaranteed to change for each update to the file, whereas time_modify is guaranteed to change only at the granularity of the time_delta attribute. Use by the client's data cache validation logic of time_modify and not change runs the risk of the client incorrectly marking stale data as valid. o Second, modified data must be flushed to the server before closing a file OPENed for write. This is complementary to the first rule. If the data is not flushed at CLOSE, the revalidation done after client OPENs as file is unable to achieve its purpose. The other aspect to flushing the data before close is that the data must be committed to stable storage, at the server, before the CLOSE operation is requested by the client. In the case of a server reboot or restart and a CLOSEd file, it may not be possible to retransmit the data to be written to the file. Hence, this requirement. 9.3.2. Data Caching and File Locking For those applications that choose to use file locking instead of share reservations to exclude inconsistent file access, there is an analogous set of constraints that apply to client side data caching. These rules are effective only if the file locking is used in a way that matches in an equivalent way the actual READ and WRITE operations executed. This is as opposed to file locking that is based on pure convention. For example, it is possible to manipulate a two-megabyte file by dividing the file into two one-megabyte regions and protecting access to the two regions by file locks on octets zero and one. A lock for write on octet zero of the file would represent the right to do READ and WRITE operations on the first region. A lock for write on octet one of the file would represent the right to do READ and WRITE operations on the second region. As long as all applications manipulating the file obey this convention, they will work on a local file system. However, they may not work with the NFS version 4 protocol unless clients refrain from data caching. The rules for data caching in the file locking environment are: Shepler, et al. Expires September 5, 2007 [Page 158] Internet-Draft NFSv4 Minor Version 1 March 2007 o First, when a client obtains a file lock for a particular region, the data cache corresponding to that region (if any cache data exists) must be revalidated. If the change attribute indicates that the file may have been updated since the cached data was obtained, the client must flush or invalidate the cached data for the newly locked region. A client might choose to invalidate all of non-modified cached data that it has for the file but the only requirement for correct operation is to invalidate all of the data in the newly locked region. o Second, before releasing a write lock for a region, all modified data for that region must be flushed to the server. The modified data must also be written to stable storage. Note that flushing data to the server and the invalidation of cached data must reflect the actual octet ranges locked or unlocked. Rounding these up or down to reflect client cache block boundaries will cause problems if not carefully done. For example, writing a modified block when only half of that block is within an area being unlocked may cause invalid modification to the region outside the unlocked area. This, in turn, may be part of a region locked by another client. Clients can avoid this situation by synchronously performing portions of write operations that overlap that portion (initial or final) that is not a full block. Similarly, invalidating a locked area which is not an integral number of full buffer blocks would require the client to read one or two partial blocks from the server if the revalidation procedure shows that the data which the client possesses may not be valid. The data that is written to the server as a prerequisite to the unlocking of a region must be written, at the server, to stable storage. The client may accomplish this either with synchronous writes or by following asynchronous writes with a COMMIT operation. This is required because retransmission of the modified data after a server reboot might conflict with a lock held by another client. A client implementation may choose to accommodate applications which use record locking in non-standard ways (e.g. using a record lock as a global semaphore) by flushing to the server more data upon an LOCKU than is covered by the locked range. This may include modified data within files other than the one for which the unlocks are being done. In such cases, the client must not interfere with applications whose READs and WRITEs are being done only within the bounds of record locks which the application holds. For example, an application locks a single octet of a file and proceeds to write that single octet. A client that chose to handle a LOCKU by flushing all modified data to the server could validly write that single octet in response to an unrelated unlock. However, it would not be valid to write the entire Shepler, et al. Expires September 5, 2007 [Page 159] Internet-Draft NFSv4 Minor Version 1 March 2007 block in which that single written octet was located since it includes an area that is not locked and might be locked by another client. Client implementations can avoid this problem by dividing files with modified data into those for which all modifications are done to areas covered by an appropriate record lock and those for which there are modifications not covered by a record lock. Any writes done for the former class of files must not include areas not locked and thus not modified on the client. 9.3.3. Data Caching and Mandatory File Locking Client side data caching needs to respect mandatory file locking when it is in effect. The presence of mandatory file locking for a given file is indicated when the client gets back NFS4ERR_LOCKED from a READ or WRITE on a file it has an appropriate share reservation for. When mandatory locking is in effect for a file, the client must check for an appropriate file lock for data being read or written. If a lock exists for the range being read or written, the client may satisfy the request using the client's validated cache. If an appropriate file lock is not held for the range of the read or write, the read or write request must not be satisfied by the client's cache and the request must be sent to the server for processing. When a read or write request partially overlaps a locked region, the request should be subdivided into multiple pieces with each region (locked or not) treated appropriately. 9.3.4. Data Caching and File Identity When clients cache data, the file data needs to be organized according to the file system object to which the data belongs. For NFS version 3 clients, the typical practice has been to assume for the purpose of caching that distinct filehandles represent distinct file system objects. The client then has the choice to organize and maintain the data cache on this basis. In the NFS version 4 protocol, there is now the possibility to have significant deviations from a "one filehandle per object" model because a filehandle may be constructed on the basis of the object's pathname. Therefore, clients need a reliable method to determine if two filehandles designate the same file system object. If clients were simply to assume that all distinct filehandles denote distinct objects and proceed to do data caching on this basis, caching inconsistencies would arise between the distinct client side objects which mapped to the same server side object. By providing a method to differentiate filehandles, the NFS version 4 protocol alleviates a potential functional regression in comparison with the NFS version 3 protocol. Without this method, caching Shepler, et al. Expires September 5, 2007 [Page 160] Internet-Draft NFSv4 Minor Version 1 March 2007 inconsistencies within the same client could occur and this has not been present in previous versions of the NFS protocol. Note that it is possible to have such inconsistencies with applications executing on multiple clients but that is not the issue being addressed here. For the purposes of data caching, the following steps allow an NFS version 4 client to determine whether two distinct filehandles denote the same server side object: o If GETATTR directed to two filehandles returns different values of the fsid attribute, then the filehandles represent distinct objects. o If GETATTR for any file with an fsid that matches the fsid of the two filehandles in question returns a unique_handles attribute with a value of TRUE, then the two objects are distinct. o If GETATTR directed to the two filehandles does not return the fileid attribute for both of the handles, then it cannot be determined whether the two objects are the same. Therefore, operations which depend on that knowledge (e.g. client side data caching) cannot be done reliably. Note that if GETATTR does not return the fileid attribute for both filehandles, it will return it for neither of the filehandles, since the fsid for both filehandles is the same. o If GETATTR directed to the two filehandles returns different values for the fileid attribute, then they are distinct objects. o Otherwise they are the same object. 9.4. Open Delegation When a file is being OPENed, the server may delegate further handling of opens and closes for that file to the opening client. Any such delegation is recallable, since the circumstances that allowed for the delegation are subject to change. In particular, the server may receive a conflicting OPEN from another client, the server must recall the delegation before deciding whether the OPEN from the other client may be granted. Making a delegation is up to the server and clients should not assume that any particular OPEN either will or will not result in an open delegation. The following is a typical set of conditions that servers might use in deciding whether OPEN should be delegated: o The client must be able to respond to the server's callback requests. The server will use the CB_NULL procedure for a test of callback ability. Shepler, et al. Expires September 5, 2007 [Page 161] Internet-Draft NFSv4 Minor Version 1 March 2007 o The client must have responded properly to previous recalls. o There must be no current open conflicting with the requested delegation. o There should be no current delegation that conflicts with the delegation being requested. o The probability of future conflicting open requests should be low based on the recent history of the file. o The existence of any server-specific semantics of OPEN/CLOSE that would make the required handling incompatible with the prescribed handling that the delegated client would apply (see below). There are two types of open delegations, read and write. A read open delegation allows a client to handle, on its own, requests to open a file for reading that do not deny read access to others. Multiple read open delegations may be outstanding simultaneously and do not conflict. A write open delegation allows the client to handle, on its own, all opens. Only one write open delegation may exist for a given file at a given time and it is inconsistent with any read open delegations. When a client has a read open delegation, it may not make any changes to the contents or attributes of the file but it is assured that no other client may do so. When a client has a write open delegation, it may modify the file data since no other client will be accessing the file's data. The client holding a write delegation may only affect file attributes which are intimately connected with the file data: size, time_modify, change. When a client has an open delegation, it does not send OPENs or CLOSEs to the server but updates the appropriate status internally. For a read open delegation, opens that cannot be handled locally (opens for write or that deny read access) must be sent to the server. When an open delegation is made, the response to the OPEN contains an open delegation structure which specifies the following: o the type of delegation (read or write) o space limitation information to control flushing of data on close (write open delegation only, see the section "Open Delegation and Data Caching") Shepler, et al. Expires September 5, 2007 [Page 162] Internet-Draft NFSv4 Minor Version 1 March 2007 o an nfsace4 specifying read and write permissions o a stateid to represent the delegation for READ and WRITE The delegation stateid is separate and distinct from the stateid for the OPEN proper. The standard stateid, unlike the delegation stateid, is associated with a particular lock_owner and will continue to be valid after the delegation is recalled and the file remains open. When a request internal to the client is made to open a file and open delegation is in effect, it will be accepted or rejected solely on the basis of the following conditions. Any requirement for other checks to be made by the delegate should result in open delegation being denied so that the checks can be made by the server itself. o The access and deny bits for the request and the file as described in the section "Share Reservations". o The read and write permissions as determined below. The nfsace4 passed with delegation can be used to avoid frequent ACCESS calls. The permission check should be as follows: o If the nfsace4 indicates that the open may be done, then it should be granted without reference to the server. o If the nfsace4 indicates that the open may not be done, then an ACCESS request must be sent to the server to obtain the definitive answer. The server may return an nfsace4 that is more restrictive than the actual ACL of the file. This includes an nfsace4 that specifies denial of all access. Note that some common practices such as mapping the traditional user "root" to the user "nobody" may make it incorrect to return the actual ACL of the file in the delegation response. The use of delegation together with various other forms of caching creates the possibility that no server authentication will ever be performed for a given user since all of the user's requests might be satisfied locally. Where the client is depending on the server for authentication, the client should be sure authentication occurs for each user by use of the ACCESS operation. This should be the case even if an ACCESS operation would not be required otherwise. As mentioned before, the server may enforce frequent authentication by returning an nfsace4 denying all access with every open delegation. Shepler, et al. Expires September 5, 2007 [Page 163] Internet-Draft NFSv4 Minor Version 1 March 2007 9.4.1. Open Delegation and Data Caching OPEN delegation allows much of the message overhead associated with the opening and closing files to be eliminated. An open when an open delegation is in effect does not require that a validation message be sent to the server. The continued endurance of the "read open delegation" provides a guarantee that no OPEN for write and thus no write has occurred. Similarly, when closing a file opened for write and if write open delegation is in effect, the data written does not have to be flushed to the server until the open delegation is recalled. The continued endurance of the open delegation provides a guarantee that no open and thus no read or write has been done by another client. For the purposes of open delegation, READs and WRITEs done without an OPEN are treated as the functional equivalents of a corresponding type of OPEN. This refers to the READs and WRITEs that use the special stateids consisting of all zero bits or all one bits. Therefore, READs or WRITEs with a special stateid done by another client will force the server to recall a write open delegation. A WRITE with a special stateid done by another client will force a recall of read open delegations. With delegations, a client is able to avoid writing data to the server when the CLOSE of a file is serviced. The file close system call is the usual point at which the client is notified of a lack of stable storage for the modified file data generated by the application. At the close, file data is written to the server and through normal accounting the server is able to determine if the available file system space for the data has been exceeded (i.e. server returns NFS4ERR_NOSPC or NFS4ERR_DQUOT). This accounting includes quotas. The introduction of delegations requires that a alternative method be in place for the same type of communication to occur between client and server. In the delegation response, the server provides either the limit of the size of the file or the number of modified blocks and associated block size. The server must ensure that the client will be able to flush data to the server of a size equal to that provided in the original delegation. The server must make this assurance for all outstanding delegations. Therefore, the server must be careful in its management of available space for new or modified data taking into account available file system space and any applicable quotas. The server can recall delegations as a result of managing the available file system space. The client should abide by the server's state space limits for delegations. If the client exceeds the stated limits for the delegation, the server's behavior is undefined. Shepler, et al. Expires September 5, 2007 [Page 164] Internet-Draft NFSv4 Minor Version 1 March 2007 Based on server conditions, quotas or available file system space, the server may grant write open delegations with very restrictive space limitations. The limitations may be defined in a way that will always force modified data to be flushed to the server on close. With respect to authentication, flushing modified data to the server after a CLOSE has occurred may be problematic. For example, the user of the application may have logged off the client and unexpired authentication credentials may not be present. In this case, the client may need to take special care to ensure that local unexpired credentials will in fact be available. This may be accomplished by tracking the expiration time of credentials and flushing data well in advance of their expiration or by making private copies of credentials to assure their availability when needed. 9.4.2. Open Delegation and File Locks When a client holds a write open delegation, lock operations are performed locally. This includes those required for mandatory file locking. This can be done since the delegation implies that there can be no conflicting locks. Similarly, all of the revalidations that would normally be associated with obtaining locks and the flushing of data associated with the releasing of locks need not be done. When a client holds a read open delegation, lock operations are not performed locally. All lock operations, including those requesting non-exclusive locks, are sent to the server for resolution. 9.4.3. Handling of CB_GETATTR The server needs to employ special handling for a GETATTR where the target is a file that has a write open delegation in effect. The reason for this is that the client holding the write delegation may have modified the data and the server needs to reflect this change to the second client that submitted the GETATTR. Therefore, the client holding the write delegation needs to be interrogated. The server will use the CB_GETATTR operation. The only attributes that the server can reliably query via CB_GETATTR are size and change. Since CB_GETATTR is being used to satisfy another client's GETATTR request, the server only needs to know if the client holding the delegation has a modified version of the file. If the client's copy of the delegated file is not modified (data or size), the server can satisfy the second client's GETATTR request from the attributes stored locally at the server. If the file is modified, the server only needs to know about this modified state. If the server determines that the file is currently modified, it will respond to Shepler, et al. Expires September 5, 2007 [Page 165] Internet-Draft NFSv4 Minor Version 1 March 2007 the second client's GETATTR as if the file had been modified locally at the server. Since the form of the change attribute is determined by the server and is opaque to the client, the client and server need to agree on a method of communicating the modified state of the file. For the size attribute, the client will report its current view of the file size. For the change attribute, the handling is more involved. For the client, the following steps will be taken when receiving a write delegation: o The value of the change attribute will be obtained from the server and cached. Let this value be represented by c. o The client will create a value greater than c that will be used for communicating modified data is held at the client. Let this value be represented by d. o When the client is queried via CB_GETATTR for the change attribute, it checks to see if it holds modified data. If the file is modified, the value d is returned for the change attribute value. If this file is not currently modified, the client returns the value c for the change attribute. For simplicity of implementation, the client MAY for each CB_GETATTR return the same value d. This is true even if, between successive CB_GETATTR operations, the client again modifies in the file's data or metadata in its cache. The client can return the same value because the only requirement is that the client be able to indicate to the server that the client holds modified data. Therefore, the value of d may always be c + 1. While the change attribute is opaque to the client in the sense that it has no idea what units of time, if any, the server is counting change with, it is not opaque in that the client has to treat it as an unsigned integer, and the server has to be able to see the results of the client's changes to that integer. Therefore, the server MUST encode the change attribute in network order when sending it to the client. The client MUST decode it from network order to its native order when receiving it and the client MUST encode it network order when sending it to the server. For this reason, change is defined as an unsigned integer rather than an opaque array of octets. For the server, the following steps will be taken when providing a write delegation: Shepler, et al. Expires September 5, 2007 [Page 166] Internet-Draft NFSv4 Minor Version 1 March 2007 o Upon providing a write delegation, the server will cache a copy of the change attribute in the data structure it uses to record the delegation. Let this value be represented by sc. o When a second client sends a GETATTR operation on the same file to the server, the server obtains the change attribute from the first client. Let this value be cc. o If the value cc is equal to sc, the file is not modified and the server returns the current values for change, time_metadata, and time_modify (for example) to the second client. o If the value cc is NOT equal to sc, the file is currently modified at the first client and most likely will be modified at the server at a future time. The server then uses its current time to construct attribute values for time_metadata and time_modify. A new value of sc, which we will call nsc, is computed by the server, such that nsc >= sc + 1. The server then returns the constructed time_metadata, time_modify, and nsc values to the requester. The server replaces sc in the delegation record with nsc. To prevent the possibility of time_modify, time_metadata, and change from appearing to go backward (which would happen if the client holding the delegation fails to write its modified data to the server before the delegation is revoked or returned), the server SHOULD update the file's metadata record with the constructed attribute values. For reasons of reasonable performance, committing the constructed attribute values to stable storage is OPTIONAL. As discussed earlier in this section, the client MAY return the same cc value on subsequent CB_GETATTR calls, even if the file was modified in the client's cache yet again between successive CB_GETATTR calls. Therefore, the server must assume that the file has been modified yet again, and MUST take care to ensure that the new nsc it constructs and returns is greater than the previous nsc it returned. An example implementation's delegation record would satisfy this mandate by including a boolean field (let us call it "modified") that is set to false when the delegation is granted, and an sc value set at the time of grant to the change attribute value. The modified field would be set to true the first time cc != sc, and would stay true until the delegation is returned or revoked. The processing for constructing nsc, time_modify, and time_metadata would use this pseudo code: Shepler, et al. Expires September 5, 2007 [Page 167] Internet-Draft NFSv4 Minor Version 1 March 2007 if (!modified) { do CB_GETATTR for change and size; if (cc != sc) modified = TRUE; } else { do CB_GETATTR for size; } if (modified) { sc = sc + 1; time_modify = time_metadata = current_time; update sc, time_modify, time_metadata into file's metadata; } return to client (that sent GETATTR) the attributes it requested, but make sure size comes from what CB_GETATTR returned. Do not update the file's metadata with the client's modified size. In the case that the file attribute size is different than the server's current value, the server treats this as a modification regardless of the value of the change attribute retrieved via CB_GETATTR and responds to the second client as in the last step. This methodology resolves issues of clock differences between client and server and other scenarios where the use of CB_GETATTR break down. It should be noted that the server is under no obligation to use CB_GETATTR and therefore the server MAY simply recall the delegation to avoid its use. 9.4.4. Recall of Open Delegation The following events necessitate recall of an open delegation: o Potentially conflicting OPEN request (or READ/WRITE done with "special" stateid) o SETATTR issued by another client o REMOVE request for the file o RENAME request for the file as either source or target of the RENAME Whether a RENAME of a directory in the path leading to the file Shepler, et al. Expires September 5, 2007 [Page 168] Internet-Draft NFSv4 Minor Version 1 March 2007 results in recall of an open delegation depends on the semantics of the server file system. If that file system denies such RENAMEs when a file is open, the recall must be performed to determine whether the file in question is, in fact, open. In addition to the situations above, the server may choose to recall open delegations at any time if resource constraints make it advisable to do so. Clients should always be prepared for the possibility of recall. When a client receives a recall for an open delegation, it needs to update state on the server before returning the delegation. These same updates must be done whenever a client chooses to return a delegation voluntarily. The following items of state need to be dealt with: o If the file associated with the delegation is no longer open and no previous CLOSE operation has been sent to the server, a CLOSE operation must be sent to the server. o If a file has other open references at the client, then OPEN operations must be sent to the server. The appropriate stateids will be provided by the server for subsequent use by the client since the delegation stateid will not longer be valid. These OPEN requests are done with the claim type of CLAIM_DELEGATE_CUR. This will allow the presentation of the delegation stateid so that the client can establish the appropriate rights to perform the OPEN. (see the section "Operation 18: OPEN" for details.) o If there are granted file locks, the corresponding LOCK operations need to be performed. This applies to the write open delegation case only. o For a write open delegation, if at the time of recall the file is not open for write, all modified data for the file must be flushed to the server. If the delegation had not existed, the client would have done this data flush before the CLOSE operation. o For a write open delegation when a file is still open at the time of recall, any modified data for the file needs to be flushed to the server. o With the write open delegation in place, it is possible that the file was truncated during the duration of the delegation. For example, the truncation could have occurred as a result of an OPEN UNCHECKED with a size attribute value of zero. Therefore, if a truncation of the file has occurred and this operation has not been propagated to the server, the truncation must occur before Shepler, et al. Expires September 5, 2007 [Page 169] Internet-Draft NFSv4 Minor Version 1 March 2007 any modified data is written to the server. In the case of write open delegation, file locking imposes some additional requirements. To precisely maintain the associated invariant, it is required to flush any modified data in any region for which a write lock was released while the write delegation was in effect. However, because the write open delegation implies no other locking by other clients, a simpler implementation is to flush all modified data for the file (as described just above) if any write lock has been released while the write open delegation was in effect. An implementation need not wait until delegation recall (or deciding to voluntarily return a delegation) to perform any of the above actions, if implementation considerations (e.g. resource availability constraints) make that desirable. Generally, however, the fact that the actual open state of the file may continue to change makes it not worthwhile to send information about opens and closes to the server, except as part of delegation return. Only in the case of closing the open that resulted in obtaining the delegation would clients be likely to do this early, since, in that case, the close once done will not be undone. Regardless of the client's choices on scheduling these actions, all must be performed before the delegation is returned, including (when applicable) the close that corresponds to the open that resulted in the delegation. These actions can be performed either in previous requests or in previous operations in the same COMPOUND request. 9.4.5. Clients that Fail to Honor Delegation Recalls A client may fail to respond to a recall for various reasons, such as a failure of the callback path from server to the client. The client may be unaware of a failure in the callback path. This lack of awareness could result in the client finding out long after the failure that its delegation has been revoked, and another client has modified the data for which the client had a delegation. This is especially a problem for the client that held a write delegation. The server also has a dilemma in that the client that fails to respond to the recall might also be sending other NFS requests, including those that renew the lease before the lease expires. Without returning an error for those lease renewing operations, the server leads the client to believe that the delegation it has is in force. This difficulty is solved by the following rules: o When the callback path is down, the server MUST NOT revoke the delegation if one of the following occurs: Shepler, et al. Expires September 5, 2007 [Page 170] Internet-Draft NFSv4 Minor Version 1 March 2007 * The client has issued a RENEW operation and the server has returned an NFS4ERR_CB_PATH_DOWN error. The server MUST renew the lease for any record locks and share reservations the client has that the server has known about (as opposed to those locks and share reservations the client has established but not yet sent to the server, due to the delegation). The server SHOULD give the client a reasonable time to return its delegations to the server before revoking the client's delegations. * The client has not issued a RENEW operation for some period of time after the server attempted to recall the delegation. This period of time MUST NOT be less than the value of the lease_time attribute. o When the client holds a delegation, it can not rely on operations, except for RENEW, that take a stateid, to renew delegation leases across callback path failures. The client that wants to keep delegations in force across callback path failures must use RENEW to do so. 9.4.6. Delegation Revocation At the point a delegation is revoked, if there are associated opens on the client, the applications holding these opens need to be notified. This notification usually occurs by returning errors for READ/WRITE operations or when a close is attempted for the open file. If no opens exist for the file at the point the delegation is revoked, then notification of the revocation is unnecessary. However, if there is modified data present at the client for the file, the user of the application should be notified. Unfortunately, it may not be possible to notify the user since active applications may not be present at the client. See the section "Revocation Recovery for Write Open Delegation" for additional details. 9.5. Data Caching and Revocation When locks and delegations are revoked, the assumptions upon which successful caching depend are no longer guaranteed. For any locks or share reservations that have been revoked, the corresponding owner needs to be notified. This notification includes applications with a file open that has a corresponding delegation which has been revoked. Cached data associated with the revocation must be removed from the client. In the case of modified data existing in the client's cache, that data must be removed from the client without it being written to the server. As mentioned, the assumptions made by the client are no longer valid at the point when a lock or delegation has been revoked. Shepler, et al. Expires September 5, 2007 [Page 171] Internet-Draft NFSv4 Minor Version 1 March 2007 For example, another client may have been granted a conflicting lock after the revocation of the lock at the first client. Therefore, the data within the lock range may have been modified by the other client. Obviously, the first client is unable to guarantee to the application what has occurred to the file in the case of revocation. Notification to a lock owner will in many cases consist of simply returning an error on the next and all subsequent READs/WRITEs to the open file or on the close. Where the methods available to a client make such notification impossible because errors for certain operations may not be returned, more drastic action such as signals or process termination may be appropriate. The justification for this is that an invariant for which an application depends on may be violated. Depending on how errors are typically treated for the client operating environment, further levels of notification including logging, console messages, and GUI pop-ups may be appropriate. 9.5.1. Revocation Recovery for Write Open Delegation Revocation recovery for a write open delegation poses the special issue of modified data in the client cache while the file is not open. In this situation, any client which does not flush modified data to the server on each close must ensure that the user receives appropriate notification of the failure as a result of the revocation. Since such situations may require human action to correct problems, notification schemes in which the appropriate user or administrator is notified may be necessary. Logging and console messages are typical examples. If there is modified data on the client, it must not be flushed normally to the server. A client may attempt to provide a copy of the file data as modified during the delegation under a different name in the file system name space to ease recovery. Note that when the client can determine that the file has not been modified by any other client, or when the client has a complete cached copy of file in question, such a saved copy of the client's view of the file may be of particular value for recovery. In other case, recovery using a copy of the file based partially on the client's cached data and partially on the server copy as modified by other clients, will be anything but straightforward, so clients may avoid saving file contents in these situations or mark the results specially to warn users of possible problems. Saving of such modified data in delegation revocation situations may be limited to files of a certain size or might be used only when sufficient disk space is available within the target file system. Such saving may also be restricted to situations when the client has Shepler, et al. Expires September 5, 2007 [Page 172] Internet-Draft NFSv4 Minor Version 1 March 2007 sufficient buffering resources to keep the cached copy available until it is properly stored to the target file system. 9.6. Attribute Caching The attributes discussed in this section do not include named attributes. Individual named attributes are analogous to files and caching of the data for these needs to be handled just as data caching is for ordinary files. Similarly, LOOKUP results from an OPENATTR directory are to be cached on the same basis as any other pathnames and similarly for directory contents. Clients may cache file attributes obtained from the server and use them to avoid subsequent GETATTR requests. Such caching is write through in that modification to file attributes is always done by means of requests to the server and should not be done locally and cached. The exception to this are modifications to attributes that are intimately connected with data caching. Therefore, extending a file by writing data to the local data cache is reflected immediately in the size as seen on the client without this change being immediately reflected on the server. Normally such changes are not propagated directly to the server but when the modified data is flushed to the server, analogous attribute changes are made on the server. When open delegation is in effect, the modified attributes may be returned to the server in the response to a CB_RECALL call. The result of local caching of attributes is that the attribute caches maintained on individual clients will not be coherent. Changes made in one order on the server may be seen in a different order on one client and in a third order on a different client. The typical file system application programming interfaces do not provide means to atomically modify or interrogate attributes for multiple files at the same time. The following rules provide an environment where the potential incoherences mentioned above can be reasonably managed. These rules are derived from the practice of previous NFS protocols. o All attributes for a given file (per-fsid attributes excepted) are cached as a unit at the client so that no non-serializability can arise within the context of a single file. o An upper time boundary is maintained on how long a client cache entry can be kept without being refreshed from the server. o When operations are performed that change attributes at the server, the updated attribute set is requested as part of the containing RPC. This includes directory operations that update Shepler, et al. Expires September 5, 2007 [Page 173] Internet-Draft NFSv4 Minor Version 1 March 2007 attributes indirectly. This is accomplished by following the modifying operation with a GETATTR operation and then using the results of the GETATTR to update the client's cached attributes. Note that if the full set of attributes to be cached is requested by READDIR, the results can be cached by the client on the same basis as attributes obtained via GETATTR. A client may validate its cached version of attributes for a file by fetching just both the change and time_access attributes and assuming that if the change attribute has the same value as it did when the attributes were cached, then no attributes other than time_access have changed. The reason why time_access is also fetched is because many servers operate in environments where the operation that updates change does not update time_access. For example, POSIX file semantics do not update access time when a file is modified by the write system call. Therefore, the client that wants a current time_access value should fetch it with change during the attribute cache validation processing and update its cached time_access. The client may maintain a cache of modified attributes for those attributes intimately connected with data of modified regular files (size, time_modify, and change). Other than those three attributes, the client MUST NOT maintain a cache of modified attributes. Instead, attribute changes are immediately sent to the server. In some operating environments, the equivalent to time_access is expected to be implicitly updated by each read of the content of the file object. If an NFS client is caching the content of a file object, whether it is a regular file, directory, or symbolic link, the client SHOULD NOT update the time_access attribute (via SETATTR or a small READ or READDIR request) on the server with each read that is satisfied from cache. The reason is that this can defeat the performance benefits of caching content, especially since an explicit SETATTR of time_access may alter the change attribute on the server. If the change attribute changes, clients that are caching the content will think the content has changed, and will re-read unmodified data from the server. Nor is the client encouraged to maintain a modified version of time_access in its cache, since this would mean that the client will either eventually have to write the access time to the server with bad performance effects, or it would never update the server's time_access, thereby resulting in a situation where an application that caches access time between a close and open of the same file observes the access time oscillating between the past and present. The time_access attribute always means the time of last access to a file by a read that was satisfied by the server. This way clients will tend to see only time_access changes that go forward in time. Shepler, et al. Expires September 5, 2007 [Page 174] Internet-Draft NFSv4 Minor Version 1 March 2007 9.7. Data and Metadata Caching and Memory Mapped Files Some operating environments include the capability for an application to map a file's content into the application's address space. Each time the application accesses a memory location that corresponds to a block that has not been loaded into the address space, a page fault occurs and the file is read (or if the block does not exist in the file, the block is allocated and then instantiated in the application's address space). As long as each memory mapped access to the file requires a page fault, the relevant attributes of the file that are used to detect access and modification (time_access, time_metadata, time_modify, and change) will be updated. However, in many operating environments, when page faults are not required these attributes will not be updated on reads or updates to the file via memory access (regardless whether the file is local file or is being access remotely). A client or server MAY fail to update attributes of a file that is being accessed via memory mapped I/O. This has several implications: o If there is an application on the server that has memory mapped a file that a client is also accessing, the client may not be able to get a consistent value of the change attribute to determine whether its cache is stale or not. A server that knows that the file is memory mapped could always pessimistically return updated values for change so as to force the application to always get the most up to date data and metadata for the file. However, due to the negative performance implications of this, such behavior is OPTIONAL. o If the memory mapped file is not being modified on the server, and instead is just being read by an application via the memory mapped interface, the client will not see an updated time_access attribute. However, in many operating environments, neither will any process running on the server. Thus NFS clients are at no disadvantage with respect to local processes. o If there is another client that is memory mapping the file, and if that client is holding a write delegation, the same set of issues as discussed in the previous two bullet items apply. So, when a server does a CB_GETATTR to a file that the client has modified in its cache, the response from CB_GETATTR will not necessarily be accurate. As discussed earlier, the client's obligation is to report that the file has been modified since the delegation was granted, not whether it has been modified again between successive CB_GETATTR calls, and the server MUST assume that any file the client has modified in cache has been modified again between successive CB_GETATTR calls. Depending on the nature of the Shepler, et al. Expires September 5, 2007 [Page 175] Internet-Draft NFSv4 Minor Version 1 March 2007 client's memory management system, this weak obligation may not be possible. A client MAY return stale information in CB_GETATTR whenever the file is memory mapped. o The mixture of memory mapping and file locking on the same file is problematic. Consider the following scenario, where a page size on each client is 8192 octets. * Client A memory maps first page (8192 octets) of file X * Client B memory maps first page (8192 octets) of file X * Client A write locks first 4096 octets * Client B write locks second 4096 octets * Client A, via a STORE instruction modifies part of its locked region. * Simultaneous to client A, client B issues a STORE on part of its locked region. Here the challenge is for each client to resynchronize to get a correct view of the first page. In many operating environments, the virtual memory management systems on each client only know a page is modified, not that a subset of the page corresponding to the respective lock regions has been modified. So it is not possible for each client to do the right thing, which is to only write to the server that portion of the page that is locked. For example, if client A simply writes out the page, and then client B writes out the page, client A's data is lost. Moreover, if mandatory locking is enabled on the file, then we have a different problem. When clients A and B issue the STORE instructions, the resulting page faults require a record lock on the entire page. Each client then tries to extend their locked range to the entire page, which results in a deadlock. Communicating the NFS4ERR_DEADLOCK error to a STORE instruction is difficult at best. If a client is locking the entire memory mapped file, there is no problem with advisory or mandatory record locking, at least until the client unlocks a region in the middle of the file. Given the above issues the following are permitted: o Clients and servers MAY deny memory mapping a file they know there are record locks for. Shepler, et al. Expires September 5, 2007 [Page 176] Internet-Draft NFSv4 Minor Version 1 March 2007 o Clients and servers MAY deny a record lock on a file they know is memory mapped. o A client MAY deny memory mapping a file that it knows requires mandatory locking for I/O. If mandatory locking is enabled after the file is opened and mapped, the client MAY deny the application further access to its mapped file. 9.8. Name Caching The results of LOOKUP and READDIR operations may be cached to avoid the cost of subsequent LOOKUP operations. Just as in the case of attribute caching, inconsistencies may arise among the various client caches. To mitigate the effects of these inconsistencies and given the context of typical file system APIs, an upper time boundary is maintained on how long a client name cache entry can be kept without verifying that the entry has not been made invalid by a directory change operation performed by another client. .LP When a client is not making changes to a directory for which there exist name cache entries, the client needs to periodically fetch attributes for that directory to ensure that it is not being modified. After determining that no modification has occurred, the expiration time for the associated name cache entries may be updated to be the current time plus the name cache staleness bound. When a client is making changes to a given directory, it needs to determine whether there have been changes made to the directory by other clients. It does this by using the change attribute as reported before and after the directory operation in the associated change_info4 value returned for the operation. The server is able to communicate to the client whether the change_info4 data is provided atomically with respect to the directory operation. If the change values are provided atomically, the client is then able to compare the pre-operation change value with the change value in the client's name cache. If the comparison indicates that the directory was updated by another client, the name cache associated with the modified directory is purged from the client. If the comparison indicates no modification, the name cache can be updated on the client to reflect the directory operation and the associated timeout extended. The post-operation change value needs to be saved as the basis for future change_info4 comparisons. As demonstrated by the scenario above, name caching requires that the client revalidate name cache data by inspecting the change attribute of a directory at the point when the name cache item was cached. This requires that the server update the change attribute for directories when the contents of the corresponding directory is modified. For a client to use the change_info4 information Shepler, et al. Expires September 5, 2007 [Page 177] Internet-Draft NFSv4 Minor Version 1 March 2007 appropriately and correctly, the server must report the pre and post operation change attribute values atomically. When the server is unable to report the before and after values atomically with respect to the directory operation, the server must indicate that fact in the change_info4 return value. When the information is not atomically reported, the client should not assume that other clients have not changed the directory. 9.9. Directory Caching The results of READDIR operations may be used to avoid subsequent READDIR operations. Just as in the cases of attribute and name caching, inconsistencies may arise among the various client caches. To mitigate the effects of these inconsistencies, and given the context of typical file system APIs, the following rules should be followed: o Cached READDIR information for a directory which is not obtained in a single READDIR operation must always be a consistent snapshot of directory contents. This is determined by using a GETATTR before the first READDIR and after the last of READDIR that contributes to the cache. o An upper time boundary is maintained to indicate the length of time a directory cache entry is considered valid before the client must revalidate the cached information. The revalidation technique parallels that discussed in the case of name caching. When the client is not changing the directory in question, checking the change attribute of the directory with GETATTR is adequate. The lifetime of the cache entry can be extended at these checkpoints. When a client is modifying the directory, the client needs to use the change_info4 data to determine whether there are other clients modifying the directory. If it is determined that no other client modifications are occurring, the client may update its directory cache to reflect its own changes. As demonstrated previously, directory caching requires that the client revalidate directory cache data by inspecting the change attribute of a directory at the point when the directory was cached. This requires that the server update the change attribute for directories when the contents of the corresponding directory is modified. For a client to use the change_info4 information appropriately and correctly, the server must report the pre and post operation change attribute values atomically. When the server is unable to report the before and after values atomically with respect to the directory operation, the server must indicate that fact in the change_info4 return value. When the information is not atomically Shepler, et al. Expires September 5, 2007 [Page 178] Internet-Draft NFSv4 Minor Version 1 March 2007 reported, the client should not assume that other clients have not changed the directory. 10. Multi-Server Name Space NFSv4.1 supports attributes that allow a namespace to extend beyond the boundaries of a single server. Use of such multi-server namespaces is optional, and for many purposes, single-server namespace are perfectly acceptable. Use of multi-server namespaces can provide many advantages, however, by separating a file system's logical position in a name space from the (possibly changing) logistical and administrative considerations that result in particular file systems being located on particular servers. 10.1. Location attributes NFSv4 contains recommended attributes that allow file systems on one server to be associated with one or more instances of that file system on other servers. These attributes specify such file systems by specifying a server name (either a DNS name or an IP address) together with the path of that file system within that server's single-server name space. The fs_locations_info recommended attribute allows specification of one more file systems instance locations where the data corresponding to a given file system may be found. This attribute provides to the client, in addition to information about file system instance locations, extensive information about the various file system instance choices (e.g. priority for use, writability, currency, etc.) as well as information to help the client efficiently effect as seamless a transition as possible among multiple file system instances, when and if that should be necessary. The fs_locations recommended attribute is inherited from NFSv4.0 and only allows specification of the file system locations where the data corresponding to a given file system may be found. Servers should make this attribute available whenever fs_locations_info is supported, but client use of fs_locations_info is to be preferred. 10.2. File System Presence or Absence A given location in an NFSv4 namespace (typically but not necessarily a multi-server namespace) can have a number of file system instance locations associated with it (via the fs_locations or fs_locations_info attribute). There may also be an actual current file system at that location, accessible via normal namespace operations (e.g. LOOKUP). In this case, the file system is said to Shepler, et al. Expires September 5, 2007 [Page 179] Internet-Draft NFSv4 Minor Version 1 March 2007 be "present" at that position in the namespace and clients will typically use it, reserving use of additional locations specified via the location-related attributes to situations in which the principal location is no longer available. When there is no actual file system at the namespace location in question, the file system is said to be "absent". An absent file system contains no files or directories other than the root and any reference to it, except to access a small set of attributes useful in determining alternate locations, will result in an error, NFS4ERR_MOVED. Note that if the server ever returns NFS4ERR_MOVED (i.e. file systems may be absent), it MUST support the fs_locations attribute and SHOULD support the fs_locations_info and fs_absent attributes. While the error name suggests that we have a case of a file system which once was present, and has only become absent later, this is only one possibility. A position in the namespace may be permanently absent with the file system(s) designated by the location attributes the only realization. The name NFS4ERR_MOVED reflects an earlier, more limited conception of its function, but this error will be returned whenever the referenced file system is absent, whether it has moved or not. Except in the case of GETATTR-type operations (to be discussed later), when the current filehandle at the start of an operation is within an absent file system, that operation is not performed and the error NFS4ERR_MOVED returned, to indicate that the file system is absent on the current server. Because a GETFH cannot succeed if the current filehandle is within an absent file system, filehandles within an absent file system cannot be transferred to the client. When a client does have filehandles within an absent file system, it is the result of obtaining them when the file system was present, and having the file system become absent subsequently. It should be noted that because the check for the current filehandle being within an absent file system happens at the start of every operation, operations which change the current filehandle so that it is within an absent file system will not result in an error. This allows such combinations as PUTFH-GETATTR and LOOKUP-GETATTR to be used to get attribute information, particularly location attribute information, as discussed below. The recommended file system attribute fs_absent can used to interrogate the present/absent status of a given file system. Shepler, et al. Expires September 5, 2007 [Page 180] Internet-Draft NFSv4 Minor Version 1 March 2007 10.3. Getting Attributes for an Absent File System When a file system is absent, most attributes are not available, but it is necessary to allow the client access to the small set of attributes that are available, and most particularly those that give information about the correct current locations for this file system, fs_locations and fs_locations_info. 10.3.1. GETATTR Within an Absent File System As mentioned above, an exception is made for GETATTR in that attributes may be obtained for a filehandle within an absent file system. This exception only applies if the attribute mask contains at least one attribute bit that indicates the client is interested in a result regarding an absent file system: fs_locations, fs_locations_info, or fs_absent. If none of these attributes is requested, GETATTR will result in an NFS4ERR_MOVED error. When a GETATTR is done on an absent file system, the set of supported attributes is very limited. Many attributes, including those that are normally mandatory will not be available on an absent file system. In addition to the attributes mentioned above (fs_locations, fs_locations_info, fs_absent), the following attributes SHOULD be available on absent file systems, in the case of recommended attributes at least to the same degree that they are available on present file systems. change: This attribute is useful for absent file systems and can be helpful in summarizing to the client when any of the location- related attributes changes. fsid: This attribute should be provided so that the client can determine file system boundaries, including, in particular, the boundary between present and absent file systems. mounted_on_fileid: For objects at the top of an absent file system this attribute needs to be available. Since the fileid is one which is within the present parent file system, there should be no need to reference the absent file system to provide this information. Other attributes SHOULD NOT be made available for absent file systems, even when it is possible to provide them. The server should not assume that more information is always better and should avoid gratuitously providing additional information. When a GETATTR operation includes a bit mask for one of the attributes fs_locations, fs_locations_info, or absent, but where the Shepler, et al. Expires September 5, 2007 [Page 181] Internet-Draft NFSv4 Minor Version 1 March 2007 bit mask includes attributes which are not supported, GETATTR will not return an error, but will return the mask of the actual attributes supported with the results. Handling of VERIFY/NVERIFY is similar to GETATTR in that if the attribute mask does not include fs_locations, fs_locations_info, or fs_absent, the error NFS4ERR_MOVED will result. It differs in that any appearance in the attribute mask of an attribute not supported for an absent file system (and note that this will include some normally mandatory attributes), will also cause an NFS4ERR_MOVED result. 10.3.2. READDIR and Absent File Systems A READDIR performed when the current filehandle is within an absent file system will result in an NFS4ERR_MOVED error, since, unlike the case of GETATTR, no such exception is made for READDIR. Attributes for an absent file system may be fetched via a READDIR for a directory in a present file system, when that directory contains the root directories of one or more absent file systems. In this case, the handling is as follows: o If the attribute set requested includes one of the attributes fs_locations, fs_locations_info, or fs_absent, then fetching of attributes proceeds normally and no NFS4ERR_MOVED indication is returned, even when the rdattr_error attribute is requested. o If the attribute set requested does not include one of the attributes fs_locations, fs_locations_info, or fs_absent, then if the rdattr_error attribute is requested, each directory entry for the root of an absent file system, will report NFS4ERR_MOVED as the value of the rdattr_error attribute. o If the attribute set requested does not include any of the attributes fs_locations, fs_locations_info, fs_absent, or rdattr_error then the occurrence of the root of an absent file system within the directory will result in the READDIR failing with an NFSERR_MOVED error. o The unavailability of an attribute because of a file system's absence, even one that is ordinarily mandatory, does not result in any error indication. The set of attributes returned for the root directory of the absent file system in that case is simply restricted to those actually available. Shepler, et al. Expires September 5, 2007 [Page 182] Internet-Draft NFSv4 Minor Version 1 March 2007 10.4. Uses of Location Information The location-bearing attributes (fs_locations and fs_locations_info), provide, together with the possibility of absent file systems, a number of important facilities in providing reliable, manageable, and scalable data access. When a file system is present, these attribute can provide alternative locations, to be used to access the same data, in the event that server failures, communications problems, or other difficulties, make continued access to the current file system impossible or otherwise impractical. Under some circumstances multiple alternative locations may be used simultaneously to provide higher performance access to the file system in question. Provision of such alternate locations is referred to as "replication" although there are cases in which replicated sets of data are not in fact present, and the replicas are instead different paths to the same data. When a file system is present and becomes absent, clients can be given the opportunity to have continued access to their data, at an alternate location. In this case, a continued attempt to use the data in the now-absent file system will result in an NFSERR_MOVED error and at that point the successor locations (typically only one but multiple choices are possible) can be fetched and used to continue access. Transfer of the file system contents to the new location is referred to as "migration", but it should be kept in mind that there are cases in which this term can be used, like "replication", when there is no actual data migration per se. Where a file system was not previously present, specification of file system location provides a means by which file systems located on one server can be associated with a name space defined by another server, thus allowing a general multi-server namespace facility. Designation of such a location, in place of an absent file system, is called "referral". 10.4.1. File System Replication The fs_locations and fs_locations_info attributes provide alternative locations, to be used to access data in place of or in a addition to the current file system instance. On first access to a file system, the client should obtain the value of the set alternate locations by interrogating the fs_locations or fs_locations_info attribute, with the latter being preferred. In the event that server failures, communications problems, or other difficulties, make continued access to the current file system Shepler, et al. Expires September 5, 2007 [Page 183] Internet-Draft NFSv4 Minor Version 1 March 2007 impossible or otherwise impractical, the client can use the alternate locations as a way to get continued access to his data. Depending on specific attributes of these alternate locations, as indicated within the fs_locations_info attribute, multiple locations may be used simultaneously, to provide higher performance through the exploitation of multiple paths between client and target file system. The alternate locations may be physical replicas of the (typically read-only) file system data, or they may reflect alternate paths to the same server or provide for the use of various form of server clustering in which multiple servers provide alternate ways of accessing the same physical file system. How these different modes of file system transition are represented within the fs_locations and fs_locations_info attributes and how the client deals with file system transition issues will be discussed in detail below. When multiple server addresses correspond to the same actual server, as shown by a common so_major_id field within the eir_server_owner field returned by EXCHANGE_ID, the client may assume that for each file system in the namespace of a given server network address, there exist file systems at corresponding namespace locations for each of the other server network addresses, even in the absence of explicit listing in fs_locations and fs_locations_info. Such corresponding file system locations can be used as alternate locations, just as those explicitly specified via the fs_locations and fs_locations_info attributes. Where these specific locations are designated in the fs_locations_info attribute, the conditions of use specified in this attribute (e.g. priorities, specification of simultaneous use) may limit the clients use of these alternate locations. When multiple replicas exist and are used simultaneously or in succession by a client, they must designate the same data (with metadata being the same to the degree indicated by the fs_locations_info attribute). Where file systems are writable, a change made on one instance must be visible on all instances, immediately upon the earlier of the return of the modifying request or the visibility of that change on any of the associated replicas. Where a file system is not writable but represents a read-only copy (possibly periodically updated) of a writable file system, similar requirements apply to the propagation of updates. It must be guaranteed that any change visible on the original file system instance must be immediately visible on any replica before the client transitions access to that replica, to avoid any possibility, that a client in effecting a transition to a replica, will see any reversion in file system state. The specific means by which this will be prevented varies based on fs4_status_type reported as part of the fs_status attribute. (See Section 10.11). Shepler, et al. Expires September 5, 2007 [Page 184] Internet-Draft NFSv4 Minor Version 1 March 2007 10.4.2. File System Migration When a file system is present and becomes absent, clients can be given the opportunity to have continued access to their data, at an alternate location, as specified by the fs_locations or fs_locations_info attribute. Typically, a client will be accessing the file system in question, get an NFS4ERR_MOVED error, and then use the fs_locations or fs_locations_info attribute to determine the new location of the data. When fs_locations_info is used, additional information will be available which will define the nature of the client's handling of the transition to a new server. Such migration can be helpful in providing load balancing or general resource reallocation. The protocol does not specify how the file system will be moved between servers. It is anticipated that a number of different server-to-server transfer mechanisms might be used with the choice left to the server implementer. The NFSv4.1 protocol specifies the method used to communicate the migration event between client and server. The new location may be an alternate communication path to the same server, or, in the case of various forms of server clustering, another server providing access to the same physical file system. The client's responsibilities in dealing with this transition depend on the specific nature of the new access path and how and whether data was in fact migrated. These issues will be discussed in detail below. When multiple server addresses correspond to the same actual server, as shown by a common value for so_major_id field of the eir_server_owner field returned by EXCHANGE_ID, the location or locations may designate alternate server addresses in the form of specific server network addresses, when the file system in question is available at those addresses, and no longer accessible at the original address. Although a single successor location is typical, multiple locations may be provided, together with information that allows priority among the choices to be indicated, via information in the fs_locations_info attribute. Where suitable clustering mechanisms make it possible to provide multiple identical file systems or paths to them, this allows the client the opportunity to deal with any resource or communications issues that might limit data availability. When an alternate location is designated as the target for migration, it must designate the same data (with metadata being the same to the degree indicated by the fs_locations_info attribute). Where file systems are writable, a change made on the original file system must Shepler, et al. Expires September 5, 2007 [Page 185] Internet-Draft NFSv4 Minor Version 1 March 2007 be visible on all migration targets. Where a file system is not writable but represents a read-only copy (possibly periodically updated) of a writable file system, similar requirements apply to the propagation of updates. Any change visible in the original file system must already be effected on all migration targets, to avoid any possibility, that a client in effecting a transition to the migration target will see any reversion in file system state. 10.4.3. Referrals Referrals provide a way of placing a file system in a location essentially without respect to its physical location on a given server. This allows a single server of a set of servers to present a multi-server namespace that encompasses file systems located on multiple servers. Some likely uses of this include establishment of site-wide or organization-wide namespaces, or even knitting such together into a truly global namespace. Referrals occur when a client determines, upon first referencing a position in the current namespace, that it is part of a new file system and that that file system is absent. When this occurs, typically by receiving the error NFS4ERR_MOVED, the actual location or locations of the file system can be determined by fetching the fs_locations or fs_locations_info attribute. The locations-related attribute may designate a single file system location or multiple file system locations, to be selected based on the needs of the client. The server, in the fs_locations_info attribute may specify priorities to be associated with various file system location choices. The server may assign different priorities to different locations as reported to individual clients, in order to adapt to client physical location or to effect load balancing. When both read-only and read-write file systems are present, some of the read-only locations may not absolutely up-to-date (as they would have to be in the case of replication and migration). Servers may also specify file system locations that include client-substituted variable so that different clients are referred to different file systems (with different data contents) based on client attributes such as cpu architecture. Use of multi-server namespaces is enabled by NFSv4 but is not required. The use of multi-server namespaces and their scope will depend on the applications used, and system administration preferences. Multi-server namespaces can be established by a single server providing a large set of referrals to all of the included file systems. Alternatively, a single multi-server namespace may be Shepler, et al. Expires September 5, 2007 [Page 186] Internet-Draft NFSv4 Minor Version 1 March 2007 administratively segmented with separate referral file systems (on separate servers) for each separately-administered section of the name space. Any segment or the top-level referral file system may use replicated referral file systems for higher availability. Generally, multi-server namespaces are for the most part uniform, in that the same data made available to one client at a given location in the namespace is made availably to all clients at that location. There are however facilities provided which allow different client to be directed to different sets of data, so as to adapt to such client characteristics as cpu architecture. 10.5. Additional Client-side Considerations When clients make use of servers that implement referrals, replication, and migration, care should be taken so that a user who mounts a given file system that includes a referral or a relocated file system continue to see a coherent picture of that user-side file system despite the fact that it contains a number of server-side file systems which may be on different servers. One important issue is upward navigation from the root of a server- side file system to its parent (specified as ".." in UNIX). The client needs to determine when it hits an fsid root going up the file tree. When at such a point, and needs to ascend to the parent, it must do so locally instead of sending a LOOKUPP call to the server. The LOOKUPP would normally return the ancestor of the target file system on the target server, which may not be part of the space that the client mounted. A related issue is upward navigation from named attribute directories. The named attribute directories are essentially detached from the namespace and this property should be safely represented in the client operating environment. LOOKUPP on a named attribute directory may return the filehandle of the associated file and conveying this to applications might be unsafe as many applications expect the parent of a directory to be a directory by itself. Therefore the client may want to hide the parent of named attribute directories (represented as ".." in UNIX) or represent the named attribute directory as its own parent (as typically done for the file system root directory in UNIX) Another issue concerns refresh of referral locations. When referrals are used extensively, they may change as server configurations change. It is expected that clients will cache information related to traversing referrals so that future client side requests are resolved locally without server communication. This is usually rooted in client-side name lookup caching. Clients should Shepler, et al. Expires September 5, 2007 [Page 187] Internet-Draft NFSv4 Minor Version 1 March 2007 periodically purge this data for referral points in order to detect changes in location information. When the change attribute changes for directories that hold referral entries or for the referral entries themselves, clients should consider any associated cached referral information to be out of date. 10.6. Effecting File System Transitions Transitions between file system instances, whether due to switching between replicas upon server unavailability, or in response to a server-initiated migration events are best dealt with together. Even though the prototypical use cases of replication and migration contain distinctive sets of features, when all possibilities for these operations are considered, the underlying unity of these operations, from the client's point of view is clear, even though for the server pragmatic considerations will normally force different implementation strategies for planned and unplanned transitions. A number of methods are possible for servers to replicate data and to track client state in order to allow clients to transition between file system instances with a minimum of disruption. Such methods vary between those that use inter-server clustering techniques to limit the changes seen by the client, to those that are less aggressive, use more standard methods of replicating data, and impose a greater burden on the client to adapt to the transition. The NFSv4.1 protocol does not impose choices on clients and servers with regard to that spectrum of transition methods. In fact, there are many valid choices, depending on client and application requirements and their interaction with server implementation choices. The NFSv4.1 protocol does define the specific choices that can be made, how these choices are communicated to the client and how the client is to deal with any discontinuities. In the sections below, references will be made to various possible server implementation choices as a way of illustrating the transition scenarios that clients may deal with. The intent here is not to define or limit server implementations but rather to illustrate the range of issues that clients may face. In the discussion below, references will be made to a file system having a particular property or of two file systems (typically the source and destination) belonging to a common class of any of several types. Two file systems that belong to such a class share some important aspect of file system behavior that clients may depend upon when present, to easily effect a seamless transition between file system instances. Conversely, where the file systems do not belong to such a common class, the client has to deal with various sorts of Shepler, et al. Expires September 5, 2007 [Page 188] Internet-Draft NFSv4 Minor Version 1 March 2007 implementation discontinuities which may cause performance or other issues in effecting a transition. Where the fs_locations_info attribute is available, such file system classification data will be made directly available to the client. See Section 10.10 for details. When only fs_locations is available, default assumptions with regard to such classifications have to be inferred. See Section 10.9 for details. In cases in which one server is expected to accept opaque values from the client that originated from another server, it is a wise implementation practice for the servers to encode the "opaque" values in big endian octet order. If this is done, servers acting as replicas or immigrating file systems will be able to parse values like stateids, directory cookies, filehandles, etc. even if their native octet order is different from that of other servers cooperating in the replication and migration of the file system. 10.6.1. File System Transitions and Simultaneous Access When a single file system may be accessed at multiple locations, whether this is because of an indication of file system identity as reported by the fs_locations or fs_locations_info attributes or because two file systems instances have corresponding locations on server addresses which connect to the same server as indicated by a common so_major_id field in the eir_server_owner field returned by EXCHANGE_ID, the client will, depending on specific circumstances as discussed below, either: o Access multiple instances simultaneously, as representing alternate paths to the same data and metadata. o The client accesses one instance (or set of instances) and then transitions to an alternative instance (or set of instances) as a result of network issues, server unresponsiveness, or server- directed migration. The transition may involve changes in filehandles, fileids, the change attribute, and or locking state, depending on the attributes of the source and destination file system instances, as specified in the fs_locations_info attribute. Which of these choices is possible, and how a transition is effected is governed by equivalence classes of file system instances as reported by the fs_locations_info attribute, and, for file systems instances in the same location within multiple single-server namespace, by the so_major_id field in the eir_server_owner field returned by EXCHANGE_ID. Shepler, et al. Expires September 5, 2007 [Page 189] Internet-Draft NFSv4 Minor Version 1 March 2007 10.6.2. Simultaneous Use and Transparent Transitions When two file system instances have the same location within their respective single-server namespaces and those two server IP addresses return the so_major_id value in the eir_server_owner value returned in response to EXCHANGE_ID, those file systems instances can be treated as the same, and either used together simultaneously or serially with no transition activity required on the part of the client. Whether simultaneous use of the two file system instances is valid is controlled by whether the fs_locations_info attribute shows the two instances as having the same _simultaneous-use_ class. Note that for two such file systems, any information within the fs_locations_info attribute that indicates the need for special transition activity, i.e. the appearance of the two file system instances with different _handle_, _fileid_, _verifier_, _change_ classes, MUST be ignored by the client. The server SHOULD not indicate that these instances belong to different _handle_, _fileid_, _verifier_, _change_ classes, whether the two instances are shown belonging to the same _simultaneous-use_ class or not. Where these conditions do not apply, a non-transparent file system instance transition is required with the details depending on the respective _handle_, _fileid_, _verifier_, _change_ classes of the two file system instances and whether the two servers in question have the same eir_server_scope value as reported by EXCHANGE_ID. 10.6.2.1. Simultaneous Use of File System Instances When the conditions above hold, in either of the following two cases, the client may use the two file system instances simultaneously. o The fs_locations_info attribute does not contain separate per-IP address entries for file systems instances at the distinct IP addresses. This includes the case in which the fs_locations_info attribute is unavailable. o The fs_locations_info attribute indicates that two file system instances belong to the same _simultaneous-use_ class. In this case, the client may use both file system instances simultaneously, as representations of the same file system, whether that happens because the two IP addresses connect to the same physical server or because different servers connect to clustered file systems and export their data in common. When simultaneous use is in effect, any change made to one file system instance must be Shepler, et al. Expires September 5, 2007 [Page 190] Internet-Draft NFSv4 Minor Version 1 March 2007 immediately reflected in the other file system instance(s). Locks are treated as part of a common lease, associated with a common client ID. Depending on the details of the eir_server_owner returned by EXCHANGE_ID, the two server instances may be accessed by different sessions or a single session in common. 10.6.2.2. Transparent File System Transitions When the conditions above hold and the fs_locations_info attribute explicitly shows the file system instances for these distinct IP addresses as belonging to different _simultaneous-use_ classes, the file system instances should not be used by the client simultaneously, but rather serially with one being used unless and until communication difficulties, lack of responsiveness, or an explicit migration event causes another file system instance (or set of file system instances sharing a common _simultaneous-use_ class to be used. When a change in file system instance is to be done, the client will use the same client ID already in effect. If it already has connections to the new server address, these will be used. Otherwise new connections to existing sessions or new sessions associated with the existing client ID are established as indicated by the eir_server_owner returned by EXCHANGE_ID. In all such transparent transition cases, the following apply: o File handles stay the same if persistent and if volatile are only subject to expiration, if they would be in the absence of file system transition. o Fileid values do not change across the transition. o The file system will have the same fsid in both the old and new locations. o Change attribute values are consistent across the transition and do not have to be refetched. When change attributes indicate that a cached object is still valid, it can remain cached. o Client, and state identifier retain their validity across the transition, except where their staleness is recognized and reported by the new server. Except where such staleness requires it, no lock reclamation is needed. o Write verifiers are presumed to retain their validity and can be presented to COMMIT, with the expectation that if COMMIT on the new server accept them as valid, then that server has all of the Shepler, et al. Expires September 5, 2007 [Page 191] Internet-Draft NFSv4 Minor Version 1 March 2007 data unstably written to the original server and has committed it to stable storage as requested. 10.6.3. Filehandles and File System Transitions There are a number of ways in which filehandles can be handled across a file system transition. These can be divided into two broad classes depending upon whether the two file systems across which the transition happens share sufficient state to effect some sort of continuity of file system handling. When there is no such co-operation in filehandle assignment, the two file systems are reported as being in different _handle_ classes. In this case, all filehandles are assumed to expire as part of the file system transition. Note that this behavior does not depend on fh_expire_type attribute and supersedes the specification of FH4_VOL_MIGRATION bit, which only affects behavior when fs_locations_info is not available. When there is co-operation in filehandle assignment, the two file systems are reported as being in the same _handle_ classes. In this case, persistent filehandle remain valid after the file system transition, while volatile filehandles (excluding those while are only volatile due to the FH4_VOL_MIGRATION bit) are subject to expiration on the target server. 10.6.4. Fileid's and File System Transitions In NFSv4.0, the issue of continuity of fileid's in the event of a file system transition was not addressed. The general expectation had been that in situations in which the two file system instances are created by a single vendor using some sort of file system image copy, fileid's will be consistent across the transition while in the analogous multi-vendor transitions they will not. This poses difficulties, especially for the client without special knowledge of the of the transition mechanisms adopted by the server. It is important to note that while clients themselves may have no trouble with a fileid changing as a result of a file system transition event, applications do typically have access to the fileid (e.g. via stat), and the result of this is that an application may work perfectly well if there is no file system instance transition or if any such transition is among instances created by a single vendor, yet be unable to deal with the situation in which a multi-vendor transition occurs, at the wrong time. Providing the same fileid's in a multi-vendor (multiple server vendors) environment has generally been held to be quite difficult. Shepler, et al. Expires September 5, 2007 [Page 192] Internet-Draft NFSv4 Minor Version 1 March 2007 While there is work to be done, it needs to be pointed out that this difficulty is partly self-imposed. Servers have typically identified fileid with inode number, i.e. with a quantity used to find the file in question. This identification poses special difficulties for migration of an fs between vendors where assigning the same index to a given file may not be possible. Note here that a fileid does not require that it be useful to find the file in question, only that it is unique within the given fs. Servers prepared to accept a fileid as a single piece of metadata and store it apart from the value used to index the file information can relatively easily maintain a fileid value across a migration event, allowing a truly transparent migration event. In any case, where servers can provide continuity of fileids, they should and the client should be able to find out that such continuity is available, and take appropriate action. Information about the continuity (or lack thereof) of fileid's across a file system is represented by specifying whether the file systems in question are of the same _fileid_ class. 10.6.5. Fsids and File System Transitions Since fsids are only unique within a per-server basis, it is to be expected that they will change during a file system transition. Clients should not make the fsid's received from the server visible to application since they may not be globally unique, and because they may change during a file system transition event. Applications are best served if they are isolated from such transitions to the extent possible. When a file system transition is made and the fs_locations_info indicates that file system in question may be split into multiple file systems (via the FSLI4F_MULTI_FS flag), client should do GETATTR's on all known objects within the file system undergoing transition, to determine the new file system boundaries. Clients may maintain the fsid's passed to existing applications by mapping all of the fsid for the descendent file systems to a the common fsid used for the original file system. 10.6.6. The Change Attribute and File System Transitions Since the change attribute is defined as a server-specific one, change attributes fetched from one server are normally presumed to be invalid on another server. Such a presumption is troublesome since it would invalidate all cached change attributes, requiring refetching. Even more disruptive, the absence of any assured continuity for the change attribute means that even if the same value is gotten on refetch no conclusions can drawn as to whether the Shepler, et al. Expires September 5, 2007 [Page 193] Internet-Draft NFSv4 Minor Version 1 March 2007 object in question has changed. The identical change attribute could be merely an artifact, of a modified file with a different change attribute construction algorithm, with that new algorithm just happening to result in an identical change value. When the two file systems have consistent change attribute formats, and this fact is communicated to the client by reporting as in the same _change_ class, the client may assume a continuity of change attribute construction and handle this situation just as it would be handled without any file system transition. 10.6.7. Lock State and File System Transitions In a file system transition, the client needs to handle cases in which the two servers have cooperated in state management and in which they have not. Cooperation by two servers in state management requires coordination of clientids. Before the client attempts to use a client ID associated with one server in a request to the server of the other file system, it must eliminate the possibility that two non-cooperating servers have assigned the same client ID by accident. The client needs to compare the eir_server_scope values returned by each server. If the scope values do not match, then the servers have not cooperated in state management. If the scope values match, then this indicates the servers have cooperated in assigning clientids to the point that they will reject clientids that refer to state they do not know about. In the case of migration, the servers involved in the migration of a file system SHOULD transfer all server state from the original to the new server. When this done, it must be done in a way that is transparent to the client. With replication, such a degree of common state is typically not the case. Clients, however should use the information provided by the eir_server_scope returned by EXCHANGE_ID to determine whether such sharing may be in effect, rather than making assumptions based on the reason for the transition. This state transfer will reduce disruption to the client when a file system transition If the servers are successful in transferring all state, the client can attempt to establish sessions associated with the client ID used for the source file system instance. If the server accepts that as a valid client ID, then the client may used the existing stateid's associated with that client ID for the old file system instance in connection with the that same client ID in connection with the file system instance. When the two servers belong to the same server scope, it does necessarily mean that when dealing with the transition, the client will not have to reclaim state. However it does mean that the client Shepler, et al. Expires September 5, 2007 [Page 194] Internet-Draft NFSv4 Minor Version 1 March 2007 may proceed using his current client ID when establishing communication with the new server and that that new server will either recognize that client ID as valid, or reject it, in which case locks must be reclaimed by the client. File systems co-operating in state management may actually share state or simply divide the id space so as to recognize (and reject as stale) each others state and clients id's. Servers which do share state may not do so under all conditions or at all times. The requirement for the server is that if it cannot be sure in accepting a client ID that it reflects the locks the client was given, it must treat all associated state as stale and report it as such to the client. When the two file systems instances are on servers that do not share a server scope value the client must establish a new client ID on the destination, if it does not have one already and reclaim if possible. In this case, old stateids and client ID's should not be presented to the new server since there is no assurance that they will not conflict with IDs valid on that server. In either case, when actual locks are not known to be maintained, the destination server may establish a grace period specific to the given file system, with non-reclaim locks being rejected for that file system, even though normal locks are being granted for other file systems. Clients should not infer the absence of a grace period for file systems being transitioned to a server from responses to requests for other file systems. In the case of lock reclamation for a given file system after a file system transition, edge conditions can arise similar to those for reclaim after server reboot (although in the case of the planned state transfer associated with migration, these can be avoided by securely recording lock state as part of state migration. Where the destination server cannot guarantee that locks will not be incorrectly granted, the destination server should not establish a file-system-specific grace period. In place of a file-system-specific version of RECLAIM_COMPLETE, servers may assume that an attempt to obtain a new lock, other than be reclaim, indicate the end of the client's attempt to reclaim locks for that file system. [NOTE: The alternative would be to adapt RECLAIM_COMPLETE to this task]. Information about client identity that may be propagated between servers in the form of client_owner4 and associated verifiers, under the assumption that the client presents the same values to all the servers with which it deals. Shepler, et al. Expires September 5, 2007 [Page 195] Internet-Draft NFSv4 Minor Version 1 March 2007 Servers are encouraged to provide facilities to allow locks to be reclaimed on the new server after a file system transition. Often, however, in cases in which the two servers do not share a server scope value, such facilities may not be available and client should be prepared to re-obtain locks, even though it is possible that the client may have his LOCK or OPEN request denied due to a conflicting lock. In some environments, such as the transition between read-only file systems, such denial of locks should not pose large difficulties in practice. When an attempt to re-establish a lock on a new server is denied, the client should treat the situation as if his original lock had been revoked. In all cases in which the lock is granted, the client cannot assume that no conflicting could have been granted in the interim. Where change attribute continuity is present, the client may check the change attribute to check for unwanted file modifications. Where even this is not available, and the file system is not read-only, a client may reasonably treat all pending locks as having been revoked. 10.6.7.1. Leases and File System Transitions In the case of lease renewal, the client may not be submitting requests for a file system that has been transferred to another server. This can occur because of the lease renewal mechanism. The client renews leases for all file systems when submitting a request on an associated session, regardless of the specific file system being referenced. In order for the client to schedule renewal of leases that may have been relocated to the new server, the client must find out about lease relocation before those leases expire. To accomplish this, the SEQUENCE operation will return the status bit SEQ4_STATUS_LEASE_MOVED, if responsibility for any of the leases to be renewed has been transferred to a new server. This condition will continue until the client receives an NFS4ERR_MOVED error and the server receives the subsequent GETATTR for the fs_locations or fs_locations_info attribute for an access to each file system for which a lease has been moved to a new server. When a client receives an SEQ4_STATUS_LEASE_MOVED indication, it should perform an operation on each file system associated with the server in question. When the client receives an NFS4ERR_MOVED error, the client can follow the normal process to obtain the new server information (through the fs_locations and fs_locations_info attributes) and perform renewal of those leases on the new server, unless information in fs_locations_info attribute shows that no state could have been transferred. If the server has not had state transferred to it transparently, the client will receive NFS4ERR_STALE_CLIENTID from the new server, as described above, and Shepler, et al. Expires September 5, 2007 [Page 196] Internet-Draft NFSv4 Minor Version 1 March 2007 the client can then reclaim locks as is done in the event of server failure. 10.6.7.2. Transitions and the Lease_time Attribute In order that the client may appropriately manage its leases in the case of a file system transition, the destination server must establish proper values for the lease_time attribute. When state is transferred transparently, that state should include the correct value of the lease_time attribute. The lease_time attribute on the destination server must never be less than that on the source since this would result in premature expiration of leases granted by the source server. Upon transitions in which state is transferred transparently, the client is under no obligation to re- fetch the lease_time attribute and may continue to use the value previously fetched (on the source server). If state has not been transferred transparently, either because the associated servers are show as have different eir_server_scope strings or because the client ID is rejected when presented to the new server, the client should fetch the value of lease_time on the new (i.e. destination) server, and use it for subsequent locking requests. However the server must respect a grace period at least as long as the lease_time on the source server, in order to ensure that clients have ample time to reclaim their lock before potentially conflicting non-reclaimed locks are granted. 10.6.8. Write Verifiers and File System Transitions In a file system transition, the two file systems may be clustered in the handling of unstably written data. When this is the case, and the two file systems belong to the same _verifier_ class, valid verifiers from one system may be recognized by the other and superfluous writes avoided. There is no requirement that all valid verifiers be recognized, but it cannot be the case that a verifier is recognized as valid when it is not. [NOTE: We need to resolve the issue of proper verifier scope]. When two file systems belong to different _verifier_ classes, the client must assume that all unstable writes in existence at the time file system transition, have been lost since there is no way the old verifier can recognized as valid (or not) on the target server. 10.7. Effecting File System Referrals Referrals are effected when an absent file system is encountered, and one or more alternate locations are made available by the Shepler, et al. Expires September 5, 2007 [Page 197] Internet-Draft NFSv4 Minor Version 1 March 2007 fs_locations or fs_locations_info attributes. The client will typically get an NFS4ERR_MOVED error, fetch the appropriate location information and proceed to access the file system on different server, even though it retains its logical position within the original namespace. The examples given in the sections below are somewhat artificial in that an actual client will not typically do a multi-component lookup, but will have cached information regarding the upper levels of the name hierarchy. However, these example are chosen to make the required behavior clear and easy to put within the scope of a small number of requests, without getting unduly into details of how specific clients might choose to cache things. 10.7.1. Referral Example (LOOKUP) Let us suppose that the following COMPOUND is issued in an environment in which /this/is/the/path is absent from the target server. This may be for a number of reasons. It may be the case that the file system has moved, or, it may be the case that the target server is functioning mainly, or solely, to refer clients to the servers on which various file systems are located. o PUTROOTFH o LOOKUP "this" o LOOKUP "is" o LOOKUP "the" o LOOKUP "path" o GETFH o GETATTR fsid,fileid,size,ctime Under the given circumstances, the following will be the result. o PUTROOTFH --> NFS_OK. The current fh is now the root of the pseudo-fs. o LOOKUP "this" --> NFS_OK. The current fh is for /this and is within the pseudo-fs. o LOOKUP "is" --> NFS_OK. The current fh is for /this/is and is within the pseudo-fs. Shepler, et al. Expires September 5, 2007 [Page 198] Internet-Draft NFSv4 Minor Version 1 March 2007 o LOOKUP "the" --> NFS_OK. The current fh is for /this/is/the and is within the pseudo-fs. o LOOKUP "path" --> NFS_OK. The current fh is for /this/is/the/path and is within a new, absent fs, but ... the client will never see the value of that fh. o GETFH --> NFS4ERR_MOVED. Fails because current fh is in an absent fs at the start of the operation and the spec makes no exception for GETFH. o GETATTR fsid,fileid,size,ctime. Not executed because the failure of the GETFH stops processing of the COMPOUND. Given the failure of the GETFH, the client has the job of determining the root of the absent file system and where to find that file system, i.e. the server and path relative to that server's root fh. Note here that in this example, the client did not obtain filehandles and attribute information (e.g. fsid) for the intermediate directories, so that he would not be sure where the absent file system starts. It could be the case, for example, that /this/is/the is the root of the moved file system and that the reason that the lookup of "path" succeeded is that the