idnits 2.17.1 draft-ietf-nfsv4-migration-issues-13.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (May 16, 2017) is 2537 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 5661 (Obsoleted by RFC 8881) Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 NFSv4 D. Noveck, Ed. 3 Internet-Draft NetApp 4 Intended status: Informational P. Shivam 5 Expires: November 17, 2017 C. Lever 6 B. Baker 7 ORACLE 8 May 16, 2017 10 NFSv4 Migration and Trunking: Implementation and Specification Issues 11 draft-ietf-nfsv4-migration-issues-13 13 Abstract 15 This document discusses a range of implementation and specification 16 issues concerning features related to the use of location-related 17 attributes in NFSv4. These include migration, which transfers 18 responsibility for a file system from one server to another, and 19 trunking which deals with the discovery and control of the set of 20 network addresses to use to access a file system. The focus of the 21 discussion, which relates to multiple minor versions, is on providing 22 appropriate clarification and correction of existing specifications. 24 Status of This Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at http://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on November 17, 2017. 41 Copyright Notice 43 Copyright (c) 2017 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (http://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 59 2. Language . . . . . . . . . . . . . . . . . . . . . . . . . . 4 60 2.1. Requirements Language . . . . . . . . . . . . . . . . . . 4 61 2.2. Use of Normative Terms . . . . . . . . . . . . . . . . . 4 62 2.3. Terminology Used in this Document . . . . . . . . . . . . 5 63 3. Multi-version Issues and Their Resolution . . . . . . . . . . 6 64 3.1. Multi-Version Issues . . . . . . . . . . . . . . . . . . 6 65 3.2. Resolution of Multi-Version Issues . . . . . . . . . . . 7 66 3.2.1. Providing Trunking Discovery . . . . . . . . . . . . 7 67 3.2.2. Interaction of Trunking and Migration . . . . . . . . 9 68 4. NFSv4.0 Issues . . . . . . . . . . . . . . . . . . . . . . . 11 69 4.1. Core NFSv4.0 Migration Issues . . . . . . . . . . . . . . 11 70 4.2. Resolution of Core Migration Protocol Difficulties in 71 NFSv4.0 . . . . . . . . . . . . . . . . . . . . . . . . . 12 72 4.3. Additional NFSv4.0 Issues . . . . . . . . . . . . . . . . 12 73 4.4. Resolution of Additional NFSv4.0 Issues . . . . . . . . . 12 74 5. Issues for NFSv4.1 . . . . . . . . . . . . . . . . . . . . . 14 75 5.1. Issues to Address for NFSv4.1 . . . . . . . . . . . . . . 14 76 5.1.1. Addressing State Merger in NFSv4.1 . . . . . . . . . 15 77 5.1.2. Addressing pNFS Relationship with Migration . . . . . 15 78 5.1.3. Addressing Server_owner Changes in NFSv4.1 . . . . . 16 79 5.1.4. Addressing Confirmation Status of Migrated 80 Client IDs in NFSv4.1 . . . . . . . . . . . . . . . . 17 81 5.1.5. Addressing Changes in Trunking Configuration . . . . 18 82 5.1.6. Addressing Session Migration in NFSv4.1 . . . . . . . 18 83 5.2. Possible Resolutions for NFSv4.1 Protocol Issues . . . . 19 84 5.2.1. Client ID Confirmation Issues . . . . . . . . . . . . 19 85 5.2.2. Dealing with Multiple Location Entries . . . . . . . 20 86 5.2.3. Migration and pNFS . . . . . . . . . . . . . . . . . 22 87 5.3. Defining Server Responsibilities for NFSv4.1 . . . . . . 23 88 5.3.1. Server Responsibilities in Effecting Transparent 89 State Migration . . . . . . . . . . . . . . . . . . . 24 90 5.3.2. Synchronizing Session Transfer . . . . . . . . . . . 25 91 5.4. Defining Client Responsibilities for NFSv4.1 . . . . . . 27 92 5.4.1. Client Recovery from Migration Events . . . . . . . . 28 93 5.4.2. The Migration Discovery Process . . . . . . . . . . . 30 94 5.4.3. Overview of Client Response to NFS4ERR_MOVED . . . . 31 95 5.4.4. Obtaining Access to Sessions and State after 96 Migration . . . . . . . . . . . . . . . . . . . . . . 33 98 5.4.5. Obtaining Access to Sessions and State after Network 99 Address Transfer . . . . . . . . . . . . . . . . . . 35 100 5.5. Resolution of NFSv4.1 Issues . . . . . . . . . . . . . . 35 101 6. Security Considerations . . . . . . . . . . . . . . . . . . . 38 102 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 38 103 8. Normative References . . . . . . . . . . . . . . . . . . . . 38 104 Appendix A. Acknowledgements . . . . . . . . . . . . . . . . . . 38 105 Appendix B. History of this Document . . . . . . . . . . . . . . 39 106 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 40 108 1. Introduction 110 This is an informational document that discusses a number of related 111 issues in multiple versions of NFSv4. 113 Many of these relate to the migration feature of NFSv4, which 114 provides for moving responsibility for a single filesystem from one 115 server to another, without disruption to clients. A number of 116 problems in the specification of this feature in NFSv4.0 were 117 resolved by the publication of [RFC7931], which added trunking 118 detection to NFSV4.0. However, NFSv4.0 remains without an 119 appropriate discussion of trunking discovery, which has many 120 important connections with migration. As a result, NFSv4.0 requires 121 clarification of how the client is to respond to changes in the 122 trunking arrangements to use, both when migration occurs and when it 123 does not. 125 In addition, there are specification issues to be resolved with 126 regard to the NFSv4.1 version of these features which are discussed 127 in this document. 129 All of the issues discussed relate to the handling and interpretation 130 of the location-related attributes fs_locations and fs_locations_info 131 and to the proper client and server handling of changes in the values 132 of these attributes 134 These issues are all related to the protocol features for effecting 135 file system migration, or to trunking discovery but it is not 136 possible to treat each of these features in isolation. These 137 features are inherently linked because migration needs to deal with 138 the possibility of multiple server addresses in location attributes 139 and because location attributes, which provide trunking-related 140 information, may change, which might or might not involve migration. 142 2. Language 144 2.1. Requirements Language 146 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 147 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 148 document are to be interpreted as described in [RFC2119]. 150 2.2. Use of Normative Terms 152 This document, which deals with existing issues/problems in 153 standards-track documents, is in the informational category, and 154 while the facts it reports may have normative implications, any such 155 normative significance reflects the readers' preferences. For 156 example, we may report that the existing definition of migration for 157 NFSv4.1 does not properly describe how migrating state is to be 158 merged with existing state for the destination server. While it is 159 to be expected that client and server implementers will judge this to 160 be a situation that it would be appropriate to resolve, the judgment 161 as to how pressing this issue should be considered is a judgment for 162 the reader, and eventually the nfsv4 working group to make. 164 We do explore possible ways in which such issues can be dealt with, 165 with minimal negative effects, given that the working group has 166 decided to address these issues, but the choice of exactly how to 167 address these is best given effect in one or more standards-track 168 documents and/or errata. 170 In the context of this informational document, these normative 171 keywords will generally occur in the context of a quotation, most 172 often direct but sometimes indirect. The context will make it clear 173 whether the quotation is from: 175 o The base definition of the NFSv4.0 protocol [RFC7530]. 177 o The document updating the handling of migration in NFSv4.0 178 [RFC7931]. 180 o The current definition of the NFSv4.1 protocol [RFC5661]. 182 An additional possibility is that these terms may appear in a 183 proposed or possible text to serve as a replacement for a current 184 protocol specification. Sometimes, a number of possible alternative 185 texts may be listed and benefits and detriments of each examined in 186 turn. 188 2.3. Terminology Used in this Document 190 In this document the phrase "client ID" always refers to the 64-bit 191 shorthand identifier assigned by the server (a clientid4) and never 192 to the structure which the client uses to identify itself to the 193 server (called an nfs_client_id4 or client_owner in NFSv4.0 and 194 NFSv4.1 respectively). The opaque identifier within those structures 195 is referred to as a client id string". 197 Regarding trunking of network addresses, we use the following 198 terminology: 200 o Trunking detection refers to ways of deciding whether two specific 201 network addresses are connected to the same NFSv4 server. The 202 means available to make this determination depends on the protocol 203 version, and, in some cases, on the client implementation. 205 o Two network addresses connected to the same server are said to be 206 server-trunkable. 208 o Two network addresses connected to the same server such that those 209 addresses can be used to support a single common session are 210 referred to as session-trunkable. Note that two addresses may be 211 server-trunkable without being session-trunkable. 213 o Trunking discovery is a process by which a client using one 214 network address can obtain other addresses that are trunkable 215 (either server-trunkable or session-trunkable) with it. 217 Regarding terminology relating to attributes used in trunking 218 discovery and other multi-server namespace features: 220 o Location attributes include the fs_locations and fs_locations_info 221 attributes. 223 o Location entries are the individual file system locations in the 224 location attributes. 226 o Location elements are derived from location entries. If a 227 location entry specifies an IP address there is only a single 228 corresponding location element. Location entries that contain a 229 host name, are resolved using DNS, and may result in one or more 230 location elements. All location elements consist of a location 231 address which is the IP address of an interface to a server and an 232 fs name which is the location of the file system within the 233 server's pseudo-fs. The fs name is empty if the server has no 234 pseudo-fs and only a single exported file system at the root 235 filehandle. 237 o Two location elements are said to be server-trunkable if they 238 specify the same fs name and the location addresses are such that 239 the location addresses are server-trunkable. 241 o Two location elements are said to be session-trunkable if they 242 specify the same fs name and the location addresses are such that 243 the location addresses are session-trunkable. 245 Each set of server-trunkable location elements defines the available 246 access paths to a particular file system. When there are multiple 247 such file systems, each of these, which contains the same data, is a 248 replica of the others. Logically, such replication is symmetric, 249 since the fs currently in use and an alternate fs are replicas of 250 each other. Often, in other documents, the term "replica" is not 251 applied to the fs currently in use, despite the fact that the 252 replication relation is inherently symmetric. 254 3. Multi-version Issues and Their Resolution 256 3.1. Multi-Version Issues 258 The source of these issues rises from a lack of clarity regarding the 259 meaning of and proper handling for location attributes that specify 260 more than a single server address. Such situations can arise as a 261 result of multiple entries in the same attribute or because a single 262 entry has a server name which, when processed by DNS, is mapped to 263 multiple server addresses. 265 Both [RFC7530] and [RFC5661] indicate that multiple addresses may be 266 present and that these addresses may be different paths to the same 267 server as well as different copies of the same data. However, the 268 following issues have, for both protocols, interfered with the 269 recognition of the existing location attributes as a way of providing 270 a trunking discovery function: 272 o There is no discussion of the use of these attributes when a file 273 system is first accessed, giving the impression that they are only 274 to be used as a way of overcoming access difficulties. 276 o The treatment of migration (and in the case of NFSv4.1 of file 277 system transitions in general) is written as if only a single 278 server address will be accessed. 280 o Although location attributes can contains the addresses of 281 migration targets and of additional replicas as well, the issues 282 that arise when both of these are specified are not clearly 283 discussed. 285 In addition, there are factors that relates to specific protocol 286 versions and documents: 288 o In NFSv4.0, as described solely by [RFC7530], trunking is treated 289 as a problem to be avoided, making the whole matter moot. 291 o In NFSv4.0, as described by [RFC7530] together with [RFC7931], the 292 situation is different. There is a means of trunking detection 293 suggested in [RFC7931] but it is a suggestion only valid when the 294 client chooses to use the uniform client id string model. 296 o For NFSv4.1, as described by [RFC5661], there is a standard method 297 of trunking detection, which can be relied upon. 299 The issues that need to be resolved for both versions are: 301 o Provision of a trunking discovery facility to allow a client to 302 find out about other addresses that may be used to access the 303 current server. 305 o Better integration of migration with trunking changes, including 306 situations in which the set of addresses to access the same server 307 changes (without migration) and those in which there is a shift to 308 a different server, but trunking of addresses on either the source 309 or destination is involved 311 3.2. Resolution of Multi-Version Issues 313 Although the specifics of addressing these issues will be different 314 for different versions, there are some common aspects discussed in 315 the subsections below: 317 o The trunking discovery function, except for the additional 318 location attribute fs_locations_info, is to be addressed in 319 substantially the same way in both versions, as explained in 320 Section 3.2.1. 322 o The interaction of trunking and migration is discussed in general 323 terms in Section 3.2.2. However, the specifics of the NFSv4.1 324 client's response to NFS4ERR_MOVED are discussed in Sections 325 5.4.3, 5.4.4, and 5.4.5. 327 3.2.1. Providing Trunking Discovery 329 A client can discover a set network addresses to use to access a file 330 system using an NFSv4 server in a number of ways: 332 o If the client is accessing a server using its name, that name can 333 be mapped to a set of IP addresses using DNS and if multiple 334 addresses are available, those addresses can generally be used 335 together to access the server. 337 o A client connected to a server without knowledge of its name can 338 obtain the value of a location attribute (e.g. fs_locations). 339 Where an entry within that attribute specifies a server name, DNS 340 can be used to obtain one or more network addresses corresponding 341 to that name. In cases in which one of those is the address being 342 used, the others that corresponding to that name can also be used 343 to access the server. 345 o A client can obtain the value of a location attribute (e.g. 346 fs_locations) and use location entries that specify network 347 addresses. When there is a means of trunking detection available 348 (see below), all of addresses that are determined to correspond to 349 the same server can be used to access the server. 351 Note that the last two of these are usable in situations in which 352 NFS4ERR_MOVED was returned. Note that this does not necessarily mean 353 that migration has occurred since there may be a shift in the set of 354 network addresses to be used without changing to a different server. 355 See Section 3.2.2 for further discussion. 357 Which of the above means of providing trunking information is 358 appropriate to use in a given environment will depend on security 359 considerations, the possible need for the server to direct different 360 clients to different sets of addresses, and the availability of 361 trunking detection facilities on the clients. 363 With regard to security, the possibility that requests to determine 364 the set of network addresses corresponding to a given server might be 365 interfered with or have their responses corrupted needs to be taken 366 account of. As a result, when use of DNSSEC is not available, it 367 might not be advisable to present server names in location attributes 368 and present the network addresses directly, eliminating the need to 369 use DNS to effect this translation. Fetching of location attributes 370 should be done with integrity protection. 372 In some situations, the server will want to direct clients to use 373 specific sets of network addresses, to effect load balancing, or to 374 meet quality-of-service goals. In such environments, presentation of 375 network addresses directly in the location attribute can help give 376 the server the necessary control over the paths to be used when 377 accessing particular file systems. When such techniques are used, 378 servers typically present their own network addresses in the location 379 attribute while adding the names of other servers, such as those used 380 to access replicas. 382 Trunking detection allows the client to determine whether two network 383 addresses can be used to access the same server. The availability of 384 trunking detection depends on the protocol version, and, in some 385 case, on client implementation choices: 387 o For NFSv4.0, a means by which it can be determined if two network 388 addresses correspond to the same server is suggested in [RFC7931]. 389 However is it is optional and only available to clients using the 390 uniform client id string approach. 392 o For NFSv4.1, the client can compare the server_owner returned in 393 the response to EXCHANGE_ID to determine if two network addresses 394 correspond to the same server. 396 As a result, direct presentation of network addresses in location 397 entries may be problematic for NFSv4.0, since some clients might not 398 have the trunking detection facilities that allow them to take 399 advantage of this information. For further discussion of issues 400 related to NFSv4.0, see Section 4.4. 402 3.2.2. Interaction of Trunking and Migration 404 When the set of network addresses designated by a location attribute 405 changes, NFS4ERR_MOVED may or may not result, and in some of the 406 cases in which it is returned migration will occur, while in others 407 there a shift in the network addresses used to access a particular 408 file system with no migration. 410 o When the list of networks addresses is a superset of that 411 previously in effect, there is no need for migration or any other 412 sort of client adjustment. Nevertheless, the client is free to 413 use an additional address if it provides another path to the same 414 server. If, on the other hand, it does not do so, the client may 415 treat it as it does a replica, to be used if the current server 416 addresses become unavailable. 418 o When the list of networks addresses is a subset of that previously 419 in effect, immediate action is not needed if the address was not 420 being used. The client should avoid using it in the future, 421 whether the address is for a replica or a potential additional 422 path to the server being used. 424 o When an address being removed is one of a number of paths to the 425 current server, the client can cease to use it but it can continue 426 to use it until NFS4ERR_MOVED is received. This is not considered 427 a migration event, unless it is the last available path to the 428 server that has become unusable. 430 When migration does occur, multiple addresses may be in use on the 431 server previous to migration and multiple addresses may be available 432 for use on the destination server. 434 With regard to the server in use, it may be that return of 435 NFS4ERR_MOVED indicates that a particular network address is no 436 longer to be used, without implying that migration of the file system 437 to a different server is needed In light of this possibility, clients 438 are best off not concluding that migration has occurred until 439 concluding that all the network addresses known to be associated with 440 the server are not usable. 442 It should be noted that the need to defer this determination is not 443 absolute. If a client is not aware of all network addresses for any 444 reason, if may conclude that migration has occurred when it has not 445 and treat a switch to a different server address as if it were a 446 migration event. This is generally harmless since the use of the 447 same server via a new address will appear as a successful Transparent 448 State Migration. 450 While significant harm will not arise from this misapprehension, it 451 can give rise to disconcerting situations. For example, if a lock 452 has been revoked during the address shift, it will appear to the 453 client as if the lock has been lost during migration, normally 454 calling for it to be recoverable via an fs-specific grace period 455 associated with the migration event. 457 With regard to the destination server, it is desirable for the client 458 to be aware of all the valid network addresses that can be used to 459 access the destination server. However, there is no need for this to 460 be done immediately. Implementations can process the additional 461 location elements in parallel with normal use of the first valid 462 location entry found to access the destination. 464 Because a location attribute may include entries relating to the 465 current server, the migration destination and possible replicas to 466 use, scanning for available network addresses could potentially be a 467 long process. The following list of helpful practices, here 468 presented as suggestions, could become RECOMMENDATIONs or 469 REQUIREMENTs in future standards-track documents 471 o Servers are well advised to place location entries that represent 472 addresses usable with the current server or a migration target 473 before those associated with replicas. 475 o A client can cease scanning for trunkable location entries once it 476 encounters one whose fs_name differs from the current fs name. 478 o A client can cease scanning for trunkable location entries once it 479 encounters a location element whose address in not server- 480 trunkable with the one it is using. 482 4. NFSv4.0 Issues 484 4.1. Core NFSv4.0 Migration Issues 486 Many of the problems seen with Transparent State Migration derived 487 from the inability of NFSv4.0servers to determine whether two client 488 IDs, issued on different servers, corresponded to the same client. 489 This difficulty derived in turn from the common practice, recommended 490 by [RFC7530], in which each client presented different client 491 identification strings to different servers, rather than presenting 492 the same identification string to all servers. 494 This practice, later referred to as the "non-uniform" client id 495 string approach, derived from concern that, since NFSv4.0 provided no 496 means to determine whether two IP addresses correspond to the server, 497 a single client connected to both might be confused by the fact that 498 state changes made via one IP address might unexpectedly affect the 499 state maintained with respect to the second IP address, thought of as 500 a separate server 502 To avoid this unexpected behavior, clients used the non-uniform 503 client id string approach. By doing so, a client connected to two 504 different servers (or to two IP addresses connected to the same 505 server) appeared to be two different servers. Since the server is 506 under the impression that two different clients are involved, state 507 changes made on each distinct IP address cannot be reflected on 508 another. 510 However, by doing things in that way, state migrated from server to 511 server cannot be referred to the actual client which generated it, 512 leading to confusion. 514 In addition to this core problem, the following issues with regard to 515 Transparent State Migration needed to be addressed: 517 o Clarification regarding the ability to merge state from different 518 leases even though their expiration times might not be precisely 519 synchronized. 521 o Clarifying the treatment of client IDs since it is not always 522 clear when clientid4 and when nfs_client_id4 was intended. 524 o Clarifying the logic of returning NFS4ERR_LEASE_MOVED. 526 o Clarifying the handling NFS4ERR_CLID_INUSE. 528 4.2. Resolution of Core Migration Protocol Difficulties in NFSv4.0 530 The client string identification issue was addressed in [RFC7931] as 531 follows: 533 o Defining both the uniform and non-uniform client id string 534 approaches as valid choices but indicating that the latter posed 535 difficulties for Transparent Stare Migration. 537 o Providing a way that clients using the uniform approach could use 538 to determine whether two IP addresses are connected to the same 539 server. 541 o Allowing clients using the uniform approach to avoid negative 542 consequences due to otherwise unexpected behavior since behavior 543 that is a consequence of known trunking relationships is not 544 unexpected. 546 o As a result, servers migrating state can be aware of the fact that 547 the same client is associated with two different items of state 548 even when that state was originally created on two different 549 servers. 551 Since all of the other issues noted in Section 4.1 were also 552 addressed by [RFC7931], publication of that document updating 553 [RFC7530] addressed all issues with Transparent State Migration in 554 NFSv4.0 known at that time. 556 4.3. Additional NFSv4.0 Issues 558 In light of the fact that a large set of migration-specific issues 559 were addressed by the publication of [RFC7931], the remaining issues 560 derive from those mentioned in Section 3.1. These include: 562 o Introducing facilities for trunking discovery. 564 o Clarifying the relationship between migration and trunking. 566 4.4. Resolution of Additional NFSv4.0 Issues 568 One possible approach to addressing these issues would entail 569 publication of an additional standards-track document updating 570 [RFC7530]. 572 Fortunately, it appears that all of the material to be updated 573 appears in Section 8 of that document, whether it concerns the 574 provision of trunking discovery or the interaction of trunking and 575 migration. It also appears that none of the material to be updated 576 is in sections updated by [RFC7931]. 578 A review of the existing Section 8 of [RFC7530], shows the following 579 sections as requiring significant attention: 581 o The existing Section 8.1 requires a considerable expansion to 582 explain the various uses of the fs_locations and the possible 583 interactions among them. 585 o The existing Section 8.4 may require substantial re-organization 586 to reflect the facts that fs_locations has multiple functions and 587 may be referenced on multiple occasions. 589 o The existing Section 8.5 follows the previous approach for NFSv4.0 590 in assuming that trunking simply cannot and should not happen. 591 For example, the last paragraph says: 593 If a single location entry designates multiple server IP 594 addresses, the client should choose a single one to use. When 595 two server addresses are designated by a single location entry 596 and they correspond to different servers, this normally 597 indicates some sort of misconfiguration, and so the client 598 should avoid using such location entries when alternatives are 599 available. When they are not, clients should pick one of the 600 IP addresses and use it, without using others that are not 601 directed to the same server. 603 As written, this seems to foreclose any use of trunking in 604 connection with migration. In retrospect, it appears that this 605 section should have been revised as part of [RFC7931], but since 606 that was not done then, the issue needs to be addressed now. 608 Overall, it appears that, in addition to the revision of Section 8.1, 609 Sections 8.4 and 8.5 need to be reorganized. One likely approach is 610 to divide the material into three main sections based on the 611 circumstances in which the attribute is accessed: 613 A section dealing with initial use and how it is trunking 614 discovery and preparation for replication. 616 A section dealing with subsequent attribute change and how it 617 affects trunking discovery and preparation for replication. 619 A section dealing with how the attribute is used to deal with 620 communication problems including receiving NFS4ERR_MOVED but also 621 including other communication difficulties driving selection of a 622 new replica. 624 5. Issues for NFSv4.1 626 5.1. Issues to Address for NFSv4.1 628 Because NFSv4.1 embraces the uniform client-string approach, as 629 advised by section 2.4 of [RFC5661], addressing migration issues is 630 simpler, in that a shift in client id string models is not required. 631 Instead, NFSv4 returns information in the EXCHANGE_ID response to 632 enable trunking relationships to be determined by the client. 634 Despite this simplification, there are substantial issues that need 635 to be dealt with: 637 o The other necessary part of addressing migration issues, providing 638 for the server's merger of leases that relate to the same client, 639 is not currently addressed by [RFC5661] and changes need to be 640 made to make it clear that state needs to be appropriately merged 641 as part of migration, to avoid multiple client IDs between a 642 client-server pair. 644 o The current discussion (in [RFC5661]), of the possibility of 645 server_owner changes is incomplete and confusing. 647 o As with NFSV4.0, the interaction of trunking with migration and 648 other aspects of multi-server namespace needs to be clarified. 650 Addressing migration in NFSv41 will also require adaptation of the 651 approaches used in [RFC7931] to the NFSv4.1 environment including: 653 o The use of EXCHANGE_ID needs to be accommodated including issues 654 associated with the expected confirmation status of client IDs 655 transferred by Transparent State Migration. 657 o The use of sessions needs to be addressed including discussion of 658 the proper use of the status bits returned by the SEQUENCE 659 operation. 661 In addition, there are a number of new features within NFSv4.1 whose 662 relationship with migration needs to be clarified. Some examples: 664 o There needs to be some clarification of how migration, and 665 particularly Transparent State Migration, should interact with 666 pNFS layouts. 668 o There are a number of issues related to the migration of sessions 669 that need to be addressed. 671 Discussion of how to resolve these issues will appear in the sections 672 below. 674 5.1.1. Addressing State Merger in NFSv4.1 676 The existing treatment of state transfer in [RFC5661], has similar 677 problems to that in [RFC7530] in that it assumes that the state for 678 multiple filesystems formerly on different servers will not be merged 679 so that it appears under a single common client ID. We've already 680 seen the reasons that this is a problem with regard to NFSv4.0. 682 Although we don't have the problems stemming from the non-uniform 683 client-string approach, there are a number of complexities in the 684 existing treatment of state management in the section entitled "Lock 685 State and File System Transitions" in [RFC5661] that make this non- 686 trivial to address: 688 o Migration is currently treated together with other sorts of 689 filesystem transitions including transitioning between replicas 690 without any NFS4ERR_MOVED errors. 692 o There is separate handling and discussion of the cases of matching 693 and non-matching server scopes. 695 o In the case of matching server scopes, the text calls for an 696 unrealistic degree of transparency, suggesting that the source and 697 destination servers need to cooperate in stateid and client ID 698 assignment. 700 o In the case of non-matching server scopes, the text does not 701 mention the possibility of the transparent migration of state at 702 all, resulting in a functional regression from NFSV4.0 704 5.1.2. Addressing pNFS Relationship with Migration 706 This is made difficult because, within the pNFS framework, migration 707 might mean any of several things: 709 o Transfer of the MDS, leaving DS's as they are. 711 This would be minimally disruptive to those using layouts but 712 would require the pNFS control protocol being used to support the 713 DS being directed to a new MDS. 715 o Transfer of a DS, leaving everything else in place. 717 Such a transfer can be handled without using migration at all. 718 The server can recall/revoke layouts, and issue new ones, as 719 appropriate. 721 o Transfer of the filesystem to a new filesystem with both MDS and 722 DS's moving. 724 In such a transfer, an entirely different set of DS's will be at 725 the target location. There may even be no pNFS support on the 726 destination filesystem at all. 728 Migration needs to support both the first and last of these models. 730 5.1.3. Addressing Server_owner Changes in NFSv4.1 732 Section 2.10.5 of [RFC5661] states the following. 734 The client should be prepared for the possibility that 735 eir_server_owner values may be different on subsequent EXCHANGE_ID 736 requests made to the same network address, as a result of various 737 sorts of reconfiguration events. When this happens and the 738 changes result in the invalidation of previously valid forms of 739 trunking, the client should cease to use those forms, either by 740 dropping connections or by adding sessions. For a discussion of 741 lock reclaim as it relates to such reconfiguration events, see 742 Section 8.4.2.1. 744 While this paragraph is literally true in that such reconfiguration 745 events can happen and clients have to deal with them, it is confusing 746 in that it can be read as suggesting that clients have to deal with 747 them without disruption, which in general is impossible. 749 A clearer alternative would be: 751 It is always possible that, as a result of various sorts of 752 reconfiguration events, eir_server_scope and eir_server_owner 753 values may be different on subsequent EXCHANGE_ID requests made to 754 the same network address. 756 In most cases such reconfiguration events will be disruptive and 757 indicate that an IP address formerly connected to one server is 758 now connected to an entirely different one. 760 Some guidelines on client handling of such situations follow: 762 o When eir_server_scope changes, the client has no assurance that 763 any id's it obtained previously (e.g. file handles) can be 764 validly used on the new server, and, even if the new server 765 accepts them, there is no assurance that this is not due to 766 accident. Thus it is best to treat all such state as lost/ 767 stale although a client may assume that the probability of 768 inadvertent acceptance is low and treat this situation as 769 within the next case. 771 o When eir_server_scope remains the same and 772 eir_server_owner.so_major_id changes, the client can use 773 filehandles it has and attempt reclaims. It may find that 774 these are now stale but if NFS4ERR_STALE is not received, he 775 can proceed to reclaim his opens. 777 o When eir_server_scope and eir_server_owner.so_major_id remain 778 the same, the client has to use the now-current values of 779 eir_server-owner.so_minor_id in deciding on appropriate forms 780 of trunking. 782 5.1.4. Addressing Confirmation Status of Migrated Client IDs in NFSv4.1 784 When a client ID is transferred between systems as a part of 785 migration, it has never been clear whether it should be considered 786 confirmed or unconfirmed on the target server. In the case in which 787 an associated session is transferred together with the client ID, it 788 is clear that the transferred client ID needs to be considered 789 confirmed, as the existence of an associated session is incompatible 790 with an unconfirmed client ID. 792 The case in which a client ID is transferred without an associated 793 session is less clear-cut, particularly since the treatment of 794 EXCHANGE_ID in [RFC5661] assumes that CREATE_SESSION is the only 795 means by which a client id may be confirmed. While this assumption 796 is valid in the absence of Transparent State Migration, 797 implementation of migration means that if this assumption is 798 maintained, it is not clear how migrated client ID s can be a 799 accommodated. If this assumption were maintained, we would have to 800 choose between the following two alternatives, regarding whether the 801 client ID to be reported as confirmed when EXCHANGE_ID is used to 802 register an already-known client_owner with the server. 804 o Report the client ID unconfirmed, because of the lack of an 805 associated session. This makes it simpler for the client to 806 determine whether there is an associated session transferred at 807 the same time. However, it is inconsistent with the fact there 808 are stateids which have been transferred with the client ID. 810 o Report the client ID as confirmed, because it was confirmed on the 811 source server and the transfer is not considered to have affected 812 that. Given the current description of EXCHANGE_ID in [RFC5661], 813 some modification in the treatment of client id confirmation is 814 called for. In particular, provision would have to be made to 815 enable the client id slot sequence id to be used by the client to 816 be determined. 818 Although the first approach makes it simpler for the client to 819 determine whether there is an associated session transferred at the 820 same time, it makes it more difficult to determine whether 821 Transparent State Migration has occurred. Section 5.1.6. 823 In any case, adjustments will be required to deal with the fact that 824 [RFC5661] currently assumes that a client id can only be confirmed by 825 issuing a CREATE_SESSION. In order to properly deal with the status 826 of migrated client ids, we have to distinguish among: 828 o The confirmation status as reported by EXCHANGE_ID. 830 o Whether the client id is considered confirmed as that term is used 831 in the many other cases in which the confirmation status of a 832 client ID affects how requests are handled. 834 o How the client is to determine the initial sequence id to be used 835 when doing operations such as CREATE_SESSION. 837 In [RFC5661] as it currently stands all of these are tied together 838 and it is not obvious how migrated client IDs could be accommodated 839 in this structure, and what changes are necessary to make this 840 possible. For more discussion of this issue, see Section 5.2.1. 842 5.1.5. Addressing Changes in Trunking Configuration 844 When the client us capable of finding out a set of network addresses 845 to use in accessing a server, it is always possible for that set to 846 change. 848 This sometimes requires that a network address previously used to 849 access a server becomes invalid for that purpose. This requires a 850 way of notifying the client and a way for the client to adapt to this 851 change by using a new set of network addresses to access the server. 852 his will involve recovery much like hat for migration although the 853 same server and file system is used throughout. 855 5.1.6. Addressing Session Migration in NFSv4.1 857 Some issues that need to be addressed regard the migration of 858 sessions, in addition to client IDs and stateids 859 o It needs to be made clearer how the client can deal with the 860 possibility that sessions might or might not be transferred as 861 part of Transparent State Migration. 863 o Rules need to be clarified regarding possible transfer of sessions 864 when either the source session is being used to access other file 865 systems on source server or there is already a session connecting 866 the client to the destination server. 868 o There needs to be more detail regarding how the protocol avoids 869 situations in which the same session is subject to concurrent 870 changes on two different servers at the same time. 872 5.2. Possible Resolutions for NFSv4.1 Protocol Issues 874 The subsections below explore some ways of dealing with clarifying 875 the protocol to address issues discussed in Section 5.1 877 5.2.1. Client ID Confirmation Issues 879 As mentioned previously [RFC5661], makes no provision for client IDs 880 that are confirmed other than through the use of CREATE_SESSION. For 881 example Section 18.35 of [RFC5661] states: 883 The client uses the EXCHANGE_ID operation to register a particular 884 client owner with the server. The client ID returned from this 885 operation will be necessary for requests that create state on the 886 server and will serve as a parent object to sessions created by 887 the client. In order to confirm the client ID it must first be 888 used, along with the returned eir_sequenceid, as arguments to 889 CREATE_SESSION. If the flag EXCHGID4_FLAG_CONFIRMED_R is set in 890 the result, eir_flags, then eir_sequenceid MUST be ignored, as it 891 has no relevancy. 893 In deciding how to address the status of migrated client IDs in the 894 case of Transparent State Migration, we should avoid giving undue 895 weight to the last sentence of the above simply because it is stated 896 in the form of a normative requirement. We should instead focus on 897 the reasons such terms (i.e. those defined by [RFC2119]) are to be 898 used, to state interoperability constraints. In this case, the 899 "MUST" applies to a conclusion based on the premise that a 900 CREATE_SESSION must have been done to assure that the client ID is 901 reliably known to the server. 903 In that light, let us consider a possible replacement, that treats 904 confirmation by means of CREATE_SESSION as one of a number of 905 possible means and avoids some the undesirable consequences of 906 adherence to the current approach, originally conceived without 907 taking state migration into account. 909 The client uses the EXCHANGE_ID operation to register a particular 910 client_owner with the server. However, when the client_owner has 911 been already been registered by other means (e.g. Transparent 912 State Migration), the client may still use EXCHANGE_ID to obtain 913 the client ID assigned previously. 915 The client ID returned from this operation will be associated with 916 the connection on which the EXHANGE_ID is received and will serve 917 as a parent object for sessions created by the client on this 918 connection or to which the connection is bound. As a result of 919 using those sessions to make requests involving the creation of 920 state, that state will become associated with the client ID 921 returned. 923 In situations in which the registration of the client_owner has 924 not occurred previously, the client ID must first be used, along 925 with the returned eir_sequenceid, in creating an associated 926 session using CREATE_SESSION. 928 If the flag EXCHGID4_FLAG_CONFIRMED_R is set in the result, 929 eir_flags, then it is an indication that the registration of the 930 client_owner has already occurred and that a further 931 CREATE_SESSION is not needed to confirm it. Of course, subsequent 932 CREATE_SESSION operations may be needed for other reasons. 934 The value eir_seqenceid is used to establish an initial sequence 935 value associate with the client ID returned. In cases in which a 936 CREATE_SESSION has already been done, there is no need for this 937 value, since sequencing of such request has already been 938 established and the client has no need for this value and will 939 ignore it 941 5.2.2. Dealing with Multiple Location Entries 943 The possibility that more than one server address may be present in 944 location attributes requires further clarification. This is 945 particularly the case, given the potential role of trunking for 946 NFSv4.1, whose connection to migration needs to be clarified. 948 The description of the location attributes in [RFC5661], while it 949 indicates that multiple address entries in these attributes may be 950 used to indicate alternate paths to the file system, does so mainly 951 in the context of replication and does so without mentioning 952 trunking. The discussion of migration does not discuss the 953 possibility of multiple location entries or trunking, which we will 954 explore here. 956 We will cover cases in which multiple addresses appear directly in 957 the attributes as well as those in which the multiple addresses 958 result because a single location entry is expanded into multiple 959 location elements using addresses provided by DNS. 961 When the set of valid location elements by which a file system may be 962 accessed changes, migration need not be involved. Some cases to 963 consider: 965 o When the set of location elements expands, migration is not 966 involved. In the case in which the additional elements are not 967 trunkable with ones previously being used, the new elements serve 968 as additional access locations, available in case of the failure 969 of server addresses being used. When additional elements are 970 trunkable with those currently being used the client may use the 971 additional addresses just as they might have if they had been 972 available when use of the file system began. 974 There is no current mechanism by which the client can be notified 975 of a change in the set of available location for an fs. Given the 976 client has at least one IP address available to access the 977 filesystem in question, periodic polling is an adequate mechanism 978 for the client to find additional server addresses to use to 979 access the file system. 981 o When the set of location elements contracts but none of the 982 elements no longer usable were in fact being used by the client, 983 then no migration is involved and no change in network addresses 984 is needed. Only if the client were to start using one of the 985 unavailable elements would the client be notified (via 986 NFS4ERR_MOVED) of the need to not use those elements and to use 987 others provided by a location attribute. 989 When a specific server address being used becomes unavailable to 990 service a particular file system, NFS4ERR_MOVED will be returned, and 991 the client will respond based on the available locations. Whether 992 continuity of locking state will be available depends on a number of 993 factors: 995 o If there are still elements in use trunkable with the element that 996 has become unavailable, there will still be a continuity of 997 locking state, even though Transparent State Migration per se has 998 not occurred. If the in-use addresses are session-trunkable with 999 the address becoming unavailable, only one connection is lost and 1000 all existing sessions will remain available. If, on the other 1001 hand, the in-use addresses are only clientid-trunkable with the 1002 address becoming unavailable, a session can be lost. However, 1003 that session can be made available on those other nodes, just as 1004 they it would have been if Transparent State Migration were in 1005 effect, even though no migration has occurred. 1007 o Otherwise, if there are available addresses trunkable with the one 1008 that has become unavailable, the client has access to existing 1009 locking state once it establishes a connection with the new 1010 addresses, using a new or existing session depending on the type 1011 of trunking in effect. This is also similar to the case in which 1012 Transparent State Migration has occurred, even though there is no 1013 migration, with the state remaining on the existing server. 1015 Note that this case, as well as the previous one, can be expected 1016 in the case in which the server seeks to direct traffic with 1017 regard to particular file systems to choose addresses, in the 1018 interest of load balancing, to adjust to hardware availability 1019 constraints, or for other reasons. 1021 o In other cases, migration has occurred and the client can 1022 determine whether Transparent State Migration occurred and whether 1023 any locking state was lost during the transfer. 1025 Whether migration has occurred or not, the client can use the 1026 procedure described in Section 5.4.3 to recover access to existing 1027 locking state and, in some cases, sessions. 1029 One should note the following differences between migration with 1030 Transparent State Migration and the similar cases in which there is a 1031 continuity of locking state with no change in the server. 1033 o When locks are lost (as indicated when using them or via the 1034 SEQ4_STATUS flags) and migration has not been done, they are not 1035 to be reclaimed, except when SEQ4_STATUS_RESTART_RECLAIM_NEEDED is 1036 set. Instead such losses are treated as lock revocations and 1037 acknowledged using FREE_STATEID. 1039 o When migration has not been done, there is no need for a 1040 RECLAIM_COMPLETE (with rca_one_fs set to true). 1042 5.2.3. Migration and pNFS 1044 When pNFS is involved, the protocol is capable of supporting: 1046 o Migration of the MDS, leaving DS's in place. 1048 o Migration of the file system as a whole, including the MDS and 1049 associated DS's. 1051 o Replacement of one DS by another. 1053 o Migration of a pNFS file system to one in which pNFS is not used. 1055 o Migration of a file system not using pNFS to one in which layouts 1056 are available. 1058 Migration of the MDS function is directly supported by Transparent 1059 State Migration. Layout state will normally be transparently 1060 transferred, just as other state is. As a result, Transparent State 1061 Migration provides a framework in which, given appropriate inter-MDS 1062 data transfer, one MDS can be substituted for another. 1064 Migration of the file system function as a whole can be accomplished 1065 by recalling all layouts as part of the initial phase of the 1066 migration process. As a result, IO will be done through the MDS 1067 during the migration process, and new layouts can be granted once the 1068 client is interacting with the new MDS. An MDS can also effect this 1069 sort of transition by revoking all layouts as part of Transparent 1070 State Migration, as long as the client is notified about the loss of 1071 state. 1073 In order to allow migration to a file system on which pNFS is not 1074 supported, clients need to be prepared for a situation in which 1075 layouts are not available or supported on the destination file system 1076 and so direct IO requests to the destination server, rather than 1077 depending on layouts being available. 1079 Replacement of one DS by another is not addressed by migration as 1080 such but can be effected by an MDS recalling layouts for the DS to be 1081 replaced and issuing new ones to be served by the successor DS. 1083 Migration may transfer a file system from a server which does not 1084 support pNFS to one which does. In order to properly adapt to this 1085 situation, clients which support pNFS, but function adequately in its 1086 absence, should check for pNFS support when a file system is migrated 1087 and be prepared to use pNFS when support is available. 1089 5.3. Defining Server Responsibilities for NFSv4.1 1091 The subsections below discuss the responsibilities of source and 1092 destination servers in effecting the necessary transfer of 1093 information to support Transparent State Migration. 1095 5.3.1. Server Responsibilities in Effecting Transparent State Migration 1097 The basic responsibility of the source server in effecting 1098 Transparent State Migration is to make available to the destination 1099 server a description of each piece of locking state associated with 1100 the file system being migrated. In addition to client id string and 1101 verifier, the source server needs to provide, for each stateid: 1103 o The stateid including the current sequence value. 1105 o The associated client ID. 1107 o The handle of the associated file. 1109 o The type of the lock, such as open, byte-range lock, delegation, 1110 layout. 1112 o For locks such as opens and byte-range locks, there will be 1113 information about the owner(s) of the lock. 1115 o For recallable/revocable lock types, the current recall status 1116 needs to be included. 1118 o For each lock type there will by type-specific information, such 1119 as share and deny modes for opens and type and byte ranges for 1120 byte-range locks and layouts. 1122 A further server responsibility concerns locks that are revoked or 1123 otherwise lost during the process of file system migration. Because 1124 locks that appear to be lost during the process of migration will be 1125 reclaimed by the client, the servers have to take steps to ensure 1126 that locks revoked soon before or soon after migration are not 1127 inadvertently allowed to be reclaimed in situations in which the 1128 continuity of lock possession cannot be assured. 1130 o For locks lost on the source but whose loss has not yet been 1131 acknowledged by the client (by using FREE_STATEID), the 1132 destination must be aware of this loss so that it can deny a 1133 request to reclaim them. 1135 o For locks lost on the destination after the state transfer but 1136 before the client's RECLAIM_COMPLTE is done, the destination 1137 server should note these and not allow them to be reclaimed. 1139 An additional responsibility of the cooperating servers concerns 1140 situations in which a stateid cannot be transferred transparently 1141 because it conflicts with an existing stateid held by the client and 1142 associated with a different file system. In this case there are two 1143 valid choices: 1145 o Treat the transfer, as in NFSv4.0, as one without Transparent 1146 State Migration. In this case, conflicting locks cannot be 1147 granted until the client does a RECLAIM_COMPLETE, after reclaiming 1148 the locks it had, with the exception of reclaims denied because 1149 they were attempts to reclaim locks that had been lost. 1151 o Implement Transparent State Migration, except for the lock with 1152 the conflicting stateid. In this case, the client will be aware 1153 of a lost lock (through the SEQ4_STATUS flags) and be allowed to 1154 reclaim it. 1156 5.3.2. Synchronizing Session Transfer 1158 When transferring state between the source and destination, the 1159 issues discussed in Section 7.2 of [RFC7931] must still be attended 1160 to. In this case, the use of NFS4ERR_DELAY is still necessary in 1161 NFSv4.1, as it was in NFSv4.0, to prevent locking state changing 1162 while it is being transferred. 1164 There are a number of important differences in the NFS4.1 context: 1166 o The absence of RELEASE_LOCKOWNER means that the one case in which 1167 an operation could not be deferred by use of NFS4ERR_DELAY no 1168 longer exists. 1170 o Sequencing of operations is no longer done using owner-based 1171 operation sequences numbers. Instead, sequencing is session- 1172 based 1174 As a result, when sessions are not transferred, the techniques 1175 discussed in [RFC7931] are adequate and will not be further 1176 discussed. 1178 When sessions are transferred, there are a number of issues that pose 1179 challenges since, 1181 o A single session may be used to access multiple file systems, not 1182 all of which are being transferred. 1184 o Requests made on a session may, even if rejected, affect the state 1185 of the session by advancing the sequence number associated with 1186 the slot used. 1188 As a result, when the filesystem state might otherwise be considered 1189 unmodifiable, the client might have any number of in-flight requests, 1190 each of which is capable of changing session state, which may be of a 1191 number of types: 1193 1. Those requests that were processed on the migrating file system, 1194 before migration began. 1196 2. Those requests which got the error NFS4ERR_DELAY because the file 1197 system being accessed was in the process of being migrated. 1199 3. Those requests which got the error NFS4ERR_MOVED because the file 1200 system being accessed had been migrated. 1202 4. Those requests that accessed the migrating file system, in order 1203 to obtain location or status information. 1205 5. Those requests that did not reference the migrating file system. 1207 It should be noted that the history of any particular slot is likely 1208 to include a number of these request classes. In the case in which a 1209 session which is migrated is used by filesystems other than the one 1210 migrated, requests of class 5 may be common and be the last request 1211 processed, for many slots. 1213 Since session state can change even after the locking state has been 1214 fixed as part of the migration process, the session state known to 1215 the client could be different from that on the destination server, 1216 which necessarily reflects the session state on the source server, at 1217 an earlier time. In deciding how to deal with this situation, it is 1218 helpful to distinguish between two sorts of behavioral consequences 1219 of the choice of initial sequence ID values. 1221 o The error NFS4ERR_SEQ_MISORDERED is returned when the sequence ID 1222 in a request is neither equal to the last one seen for the current 1223 slot nor the next greater one. 1225 In view of the difficulty of arriving at a mutually acceptable 1226 value for the correct last sequence value at the point of 1227 migration, it may be necessary for the server to show some degree 1228 of forbearance, when the sequence ID is one that would be 1229 considered unacceptable if session migration were not involved. 1231 o Returning the cached reply for a previously executed request when 1232 the sequence ID in the request matches the last value recorded for 1233 the slot. 1235 In the cases in which an error is returned and there is no 1236 possibility of any non-idempotent operation having been executed, 1237 it may not be necessary to adhere to this as strictly as might be 1238 proper if session migration were not involved. For example, the 1239 fact that the error NFS4ERR_DELAY was returned may not assist the 1240 client in any material way, while the fact that NFS4ERR_MOVED was 1241 returned by the source server may not be relevant when the request 1242 was reissued, directed to the destination server. 1244 One part of adapting to these sorts of issues would restrict 1245 enforcement of normal slot sequence enforcement semantics until the 1246 client itself, by issuing a request using a particular slot on the 1247 destination server, established the new starting sequence for that 1248 slot on the migrated session. 1250 An important issue is that the specification needs to take note of 1251 all potential COMPOUNDs, even if they might be unlikely in practice. 1252 For example, a COMPOUND is allowed to access multiple file systems 1253 and might perform non-idempotent operations in some of them before 1254 accessing a file system being migrated. Also, a COMPOUND may return 1255 considerable data in the response, before being rejected with 1256 NFS4ERR_DELAY or NFS4ERR_MOVED, and may in addition be marked as 1257 sa_cachethis. 1259 Some possibilities that need to be considered to address the issues: 1261 o Do not enforce any sequencing semantics for a particular slot 1262 until the client has established the starting sequence for that 1263 slot on the destination server. 1265 o For each slot, do not return a cached reply returning 1266 NFS4ERR_DELAY or NFS4ERR_MOVED until the client has established 1267 the starting sequence for that slot on the destination server. 1269 o Until the client has established the starting sequence for a 1270 particular slot on the destination server, do not report 1271 NFS4ERR_SEQ_MISORDERED or return a cached reply returning 1272 NFS4ERR_DELAY or NFS4ERR_MOVED, where the reply consists solely of 1273 a series of operations where the response is NFS4_OK until the 1274 final error. 1276 5.4. Defining Client Responsibilities for NFSv4.1 1278 The subsections below discuss the responsibilities of the client in 1279 dealing with transition to a new server (migration) and to use of new 1280 network addresses in accessing existing servers. 1282 5.4.1. Client Recovery from Migration Events 1284 When a file system is migrated, there a number of migration-related 1285 status indications with which clients need to deal: 1287 o If an attempt is made to use or return a filehandle within a file 1288 system that has been migrated away from the server on which it was 1289 previously available, the error NFS4ERR_MOVED is returned. 1291 This condition continues on subsequent attempts to access the file 1292 system in question. The only way the client can avoid the error 1293 is to cease accessing the filesystem in question at its old server 1294 location and access it instead on the server to which it has been 1295 migrated. 1297 o Whenever a SEQUENCE operation is sent by a client to a server 1298 which generated state held on that client which is associated with 1299 a file system that has been migrated away from the server on which 1300 it was previously available, the status bit 1301 SEQ4_STATUS_LEASE_MOVED is set in the response. 1303 This condition continues until the client acknowledges the 1304 notification by fetching a location attribute for the migrated 1305 file system. When there are multiple migrated file systems, a 1306 location attribute for each such migrated file system needs to be 1307 fetched, in order to clear the condition. Even after the 1308 condition is cleared, the client needs to respond by using the 1309 location information to access the destination server to ensure 1310 that leases are not needlessly expired. 1312 Unlike the case of NFSv4.0 in which the corresponding conditions are 1313 both errors, in NFSv4.1 the client can, and often will, receive both 1314 indications on the same request. As a result, implementations need 1315 to address the question of how to co-ordinate the necessary recovery 1316 actions when both indications arrive simultaneously. It should be 1317 noted that when the server decides whether SEQ4_STATUS_LEASE_MOVED is 1318 to be set, it has no way of knowing which file system will be 1319 referenced or whether NFS4ERR_MOVED will be returned. 1321 While it is true that, when only a single migrated file system is 1322 involved, a single set of actions will clear both indications, the 1323 possibility of multiple migrated file systems calls for an approach 1324 in which there are separate recovery actions for each indication. In 1325 general, the response to neither indication can be subsumed within 1326 the other since: 1328 o If the client were to respond only to the MOVED indication, there 1329 would be no effective client response to a situation in which a 1330 file system was not being actively accessed at the time migration 1331 occurred. As a result, leases on the destination server might be 1332 needlessly expired. 1334 o If the client were to respond only to the LEASE_MOVED indication, 1335 recovery for migrated file systems in active use could be deferred 1336 in order to accomplish recovery for others not being actively 1337 accessed. The consequences of this choice can pose particular 1338 problems when there are a large number of file systems supported 1339 by a particular server, or when it happens that some servers, 1340 after receiving migrated file systems have periods of 1341 unavailability, such as occur as a result of server reboot. This 1342 can result in recovery for actively accessed migrated file systems 1343 being unnecessarily delayed for long periods of time. 1345 Similar considerations apply to other arrangements in which one of 1346 the indications, while not ignored per se, is subsumed within a 1347 single recovery process focused on recovery for the other indication. 1349 Although clients are free to decide on their own approaches to 1350 recovery, we will explore below an approach with the following 1351 characteristics: 1353 o All instances of the MOVED indication, whether they involve 1354 migration or not, should be dealt with promptly, either by doing 1355 the necessary recovery directly, providing that it be done 1356 asynchronously, or ensuring that it is already under way. 1358 o All instances of the LEASE_MOVED indication should be dealt with 1359 asynchronously, in a migration discovery thread whose job is to 1360 clear that indication by fetching the appropriate location 1361 attribute. Because this thread will only be fetching a location 1362 attribute and the fs_status attribute for the file systems 1363 referenced by the client, it cannot receive MOVED indications. 1364 Some useful guidance regarding possible implementation of a 1365 migration discovery thread can be found in Section 5.4.2. 1367 o When a migration discovery thread happens upon a migrated file 1368 system (i.e. not present and not a referral), the thread is likely 1369 to have cleared one (out of an unknown number) of file systems 1370 whose migration needs to be responded to. The discovery thread 1371 needs to schedule the appropriate migration recovery (as described 1372 in Section 5.4.3). This is necessary to ensure that migrated file 1373 systems will be referenced on the destination server in order to 1374 avoid unnecessary lease expiration. 1376 For many of the migrated file systems discovered in this way, the 1377 client has not received any MOVED indication. In such cases, 1378 lease recovery needs to be scheduled but it should not interfere 1379 with continuation of the migration discovery function. 1381 o When a migration discovery thread receives a LEASE_MOVED 1382 indication, it takes no special action but continues its normal 1383 operation. On the other hand, if a LEASE_MOVED indication is not 1384 received, it indicates that the thread has completed its work 1385 successfully. 1387 5.4.2. The Migration Discovery Process 1389 As noted above, LEASE_MOVED indications are best dealt with in a 1390 migration discovery thread. Because of this structure, 1392 o No action needs to be taken for such indications received by the 1393 migration discovery threads, since continuation of that thread's 1394 work will address the issue. 1396 o For such indications received in other contexts, the generally 1397 appropriate response is to initiate or otherwise provide for the 1398 execution of a migration discovery thread for file systems 1399 associated with the server IP address returning the indication. 1401 o In all cases in which the appropriate migration discovery thread 1402 is running, nothing further needs to be done to respond to 1403 LEASE_MOVED indications. 1405 This leaves a potential difficulty in situations in which the 1406 migration discovery thread is near to completion but is still 1407 operating. One should not ignore a LEASE_MOVED indication if the 1408 discovery thread is not able to respond to migrated file system 1409 without additional aid. A further difficulty in addressing such 1410 situation is that a LEASE_MOVED indication may reflect the server's 1411 state at the time the SEQUENCE operation was processed, which may be 1412 different from that in effect at the time the response is received. 1414 A useful approach to this issue involves the use of separate 1415 externally-visible discovery thread states representing non- 1416 operation, normal operation, and completion/verification of migration 1417 discovery processing. 1419 Within that framework, discovery thread processing would proceed as 1420 follows. 1422 o While in the normal-operation state, the thread would fetch, for 1423 successive file systems known to the client on the server being 1424 worked on, a location attribute plus the fs_status attribute. 1426 o If the fs_status attribute indicates that the file system is a 1427 migrated one (i.e. fss_absent is true and fss_type != 1428 STATUS4_REFERRAL) and thus that it is likely that the fetch of the 1429 location attribute has cleared one the file systems contributing 1430 to the LEASE_MOVED indication. 1432 o In cases in which that happened, the thread cannot know whether 1433 the LEASE_MOVED indication has been cleared and so it enters the 1434 completion/verification state and proceeds to issue a COMPOUND to 1435 see if the LEASE_MOVED indication has been cleared. 1437 o When the discovery thread is in the completion/verification state, 1438 if others get a LEASE_MOVED indication they note this fact and it 1439 is used when the request completes, as described below. 1441 When the request used in the completion/verification state completes: 1443 o If a LEASE_MOVED indication is returned, the discovery thread 1444 resumes its normal work. 1446 o Otherwise, if there is any record that other requests saw a 1447 LEASE_MOVED indication, that record is cleared and the 1448 verification request retried. The discovery thread remains in 1449 completion/verification state. 1451 o If there has been no LEASE_MOVED indication, the work of the 1452 discovery thread is considered completed and it enters the non- 1453 operating state. 1455 5.4.3. Overview of Client Response to NFS4ERR_MOVED 1457 This section outlines a way in which a client that receives 1458 NFS4ERR_MOVED can respond by using a new server or network address if 1459 one is available. As part of that process, it will determine: 1461 o Whether the NFS4ERR_MOVED indicates migration has occurred, or 1462 whether it indicates another sort of file system transition as 1463 discussed in Section 5.2.2. 1465 o In the case of migration, whether Transparent State Migration has 1466 occurred. 1468 o Whether any state has been lost during the process of Transparent 1469 State Migration. 1471 o Whether sessions have been transferred as part of Transparent 1472 State Migration. 1474 During the first phase of this process, the client proceeds to 1475 examine location entries to find the initial network address it will 1476 use to continue access to the file system or its replacement. For 1477 each location entry that the client examines, the process consists of 1478 five steps: 1480 1. Performing an EXCHANGE_ID is directed at the location address. 1481 This operation is used to register the client-owner with the 1482 server, to obtain a client ID to be use subsequently to 1483 communicate with it, to obtain tat client ID's confirmation 1484 status and, to determine server_owner and scope for the purpose 1485 of determining if the entry is trunkable with that previously 1486 being used to access the file system (i.e. that it represents 1487 another path to the same file system and can share locking state 1488 with it). 1490 2. Making an initial determination of whether migration has 1491 occurred. The initial determination will be based on whether the 1492 EXCHANGE_ID results indicate that the current location element is 1493 server-trunkable with that used to access the file system when 1494 access was terminated by receiving NFS4ERR_MOVED. If it is, then 1495 migration has not occurred and the transition is dealt with, at 1496 least initially, as one involving continued access to the same 1497 file system on the same server through a new network address. 1499 3. Obtaining access to existing session state or creating new 1500 sessions. How this is done depends on the initial determination 1501 of whether migration has occurred and can be done as described in 1502 Section 5.4.4 in the case of migration or as described in 1503 Section 5.4.5 in the case of a network address transfer without 1504 migration. 1506 4. Verification of the trunking relationship assumed in step 2 as 1507 discussed in Section 2.10.5.1 of [RFC5661]. Although this step 1508 will generally confirm the initial determination, it is possible 1509 for verification to fail with the result that an initial 1510 determination that a network address shift (without migration) 1511 has occurred may be invalidated and migration determined to have 1512 occurred. There is no need to redo step 3 above, since it will 1513 be possible to continue use of the session established already. 1515 5. Obtaining access to existing locking state and/or reobtaining it. 1516 How this is done depends on the final determination of whether 1517 migration has occurred and can be done as described in 1518 Section 5.4.4 in the case of migration or as described in 1519 Section 5.4.5 in the case of a network address transfer without 1520 migration. 1522 Once the initial address has been determined, clients are free to 1523 apply an abbreviated process to find additional addresses trunkable 1524 with it (clients may seek session-trunkable or server trunkable 1525 addresses depending on whether they support clientid trunking). 1526 During this later phase of the process, further location entries are 1527 examined using the abbreviated procedure specified below: 1529 1. Before the EXCHANGE_ID, the fs_name field is examined and if it 1530 does not match that currently being used, the entry is ignored. 1531 otherwise, one proceeds as specified by step 1 above,. 1533 2. In the case that the network address is session-trunkable with 1534 one used previously a BIND_CONN_TO_SESSION is used to access that 1535 session using new network address. Otherwise, or if the bind 1536 operation fails, a CREATE_SESSION is done. 1538 3. The verification procedure referred to in step 4 above is used. 1539 However, if it fails, the entry is ignored and the next available 1540 entry is used. 1542 5.4.4. Obtaining Access to Sessions and State after Migration 1544 In the event that migration has occurred, the determination of 1545 whether Transparent State Migration has occurred is driven by the 1546 client ID returned by the EXCHANGE_ID and the reported confirmation 1547 status. 1549 o If the client ID is an unconfirmed client ID not previously known 1550 to the client, then Transparent State Migration has not occurred. 1552 o If the client ID is a confirmed client ID previously known to the 1553 client, then any transferred state would have been merged with an 1554 existing client ID representing the client to the destination 1555 server. In this state merger case, Transparent State Migration 1556 might or might not have occurred and a determination as to whether 1557 it has occurred is deferred until sessions are established and we 1558 are ready to begin state recovery. 1560 o If the client ID is a confirmed client ID not previously known to 1561 the client, then the client can conclude that the client ID was 1562 transferred as part of Transparent State Migration. In this 1563 transferred client ID case, Transparent State Migration has 1564 occurred although some state may have been lost. 1566 Once the client ID has been obtained, it is necessary to obtain 1567 access to sessions to continue communication with the new server. In 1568 any of the cases in which Transparent State Migration has occurred, 1569 it is possible that a session was transferred as well. To deal with 1570 that possibility, clients can, after doing the EXCHANGE_ID, issue a 1571 BIND_CONN_TO_SESSION to connect the transferred session to a 1572 connection to the new server. If that fails, it is an indication 1573 that the session was not transferred and that a new session needs to 1574 be created to take its place. 1576 In some situations, it is possible for a BIND_CONN_TO_SESSION to 1577 succeed without session migration having occurred. If state merger 1578 has taken place then the associated client ID may have already had a 1579 set of existing sessions, with it being possible that the sessionid 1580 of a given session is the same as one that might have been migrated. 1581 In that event, a BIND_CONN_TO_SESSION might succeed, even though 1582 there could have been no migration of the session with that 1583 sessionid. 1585 Once the client has determined the initial migration status, and 1586 determined that there was a shift to a new server, it needs to re- 1587 establish its lock state, if possible. To enable this to happen 1588 without loss of the guarantees normally provided by locking, the 1589 destination server needs to implement a per-fs grace period in all 1590 cases in which lock state was lost, including those in which 1591 Transparent State Migration was not implemented. 1593 Clients need to be deal with the following cases: 1595 o In the state merger case, it is possible that the server has not 1596 attempted Transparent State Migration, in which case state may 1597 have been lost without it being reflected in the SEQ4_STATUS bits. 1598 To determine whether this has happened, the client can use 1599 TEST_STATEID to check whether the stateids created on the source 1600 server are still accessible on the destination server. Once a 1601 single stateid is found to have been successfully transferred, the 1602 client can conclude that Transparent State Migration was begun and 1603 any failure to transport all of the stateids will be reflected in 1604 the SEQ4_STATUS bits. Otherwise. Transparent State Migration has 1605 not occurred. 1607 o In a case in which Transparent State Migration has not occurred, 1608 the client can use the per-fs grace period provided by the 1609 destination server to reclaim locks that were held on the source 1610 server. 1612 o In a case in which Transparent State Migration has occurred, and 1613 no lock state was lost (as shown by SEQ4_STATUS flags), no lock 1614 reclaim is necessary. 1616 o In a case in which Transparent State Migration has occurred, and 1617 some lock state was lost (as shown by SEQ4_STATUS flags), existing 1618 stateids need to be checked for validity using TEST_STATEID, and 1619 reclaim used to re-establish any that were not transferred. 1621 For all of the cases above, RECLAIM_COMPLETE with an rca_one_fs value 1622 of true should be done before normal use of the file system including 1623 obtaining new locks for the file system. This applies even if no 1624 locks were lost and needed to be reclaimed. 1626 5.4.5. Obtaining Access to Sessions and State after Network Address 1627 Transfer 1629 The case in which there is a transfer to a new network address 1630 without migration is similar to that described in Section 5.4.4 in 1631 that there is a need to obtain access to needed sessions and locking 1632 state. However, the details are simpler and will vary depending on 1633 the type of trunking between the address receiving NFS4ERR_MOVED and 1634 that to which the transfer is to be made 1636 To make a session available for use, a BIND_CONN_TO_SESSION should be 1637 used to obtain access to the session previously in use. Only if this 1638 fails, should a CREATE_SESSION be done. While this procedure mirrors 1639 that in Section 5.4.4, there is an important difference in that 1640 preservation of the session is not purely optional but depends on the 1641 type of trunking. 1643 Access to appropriate locking state should need no actions beyond 1644 access to the session. However. the SEQ4_STATUS bits should be 1645 checked for lost locking state, including the need to reclaim locks 1646 after a server reboot. 1648 5.5. Resolution of NFSv4.1 Issues 1650 One possibility is that addressing all of the NFSv4.1 issues would 1651 entail publication of a standards-track document updating [RFC5661]. 1653 Such a document would have three major elements: 1655 o A considerable expansion of the existing Section 11.4 explaining 1656 the various uses of the location attribute and the possible 1657 interactions among these various uses. This, like the 1658 corresponding replacement section for NFSv4.0 would be based on 1659 our Section 3.2 above. Information regarding the specifics of 1660 trunking discovery might appear in this section, in a new sub- 1661 section. As part of this revision, the existing Section 11.4.2 1662 would need to be revised to explain all the possible results of 1663 NFS4ERR_MOVED including migration and a possible transparent 1664 transition in which the network address changes but the server 1665 does not. 1667 o A revision of the existing section 18.35 (dealing with 1668 EXCHANGE_ID) addressing the issues discussed in Section 5.2.1. 1670 o A major replacement of the existing Section 11.7, entitled 1671 "Effecting File System Transitions", as discussed below. 1673 In addition, there is a set of smaller changes necessary 1675 o Update the existing Section 2.10.5 to clarify the proper response 1676 to server_owner changes, as described in our Section 5.1.3. 1678 o Replacement of the existing Section 15.1.2.4 to reflect the fact 1679 that NFS4ERR_MOVED can occur when a file system is now accessible 1680 at a different network address. A possible replacement text might 1681 be: 1683 The file system that contains the current filehandle object is 1684 not accessible using the network address which has been used. 1685 It may have been relocated, migrated to another server, be 1686 accessible using another network address on the current server, 1687 or it may have never been present. The client may obtain the 1688 new file system location by obtaining the "fs_locations" or 1689 "fs_locations_info" attribute for the current filehandle. For 1690 further discussion, refer to Section 11.4.2 1692 The replacement for the existing section 11.7 would maintain most 1693 sections essentially as they are, only making minor changes to 1694 include server-trunking in the discussion. However, in some cases 1695 involving more significant changes to existing sub-sections, and 1696 potential new sub-sections are listed below: 1698 o The existing Section 11.7.1 needs to be modified to refer 1699 explicitly to the previous discussion of trunking discovery. 1701 In addition, the term "multi-home single-server namespace", used 1702 nowhere else in [RFC5661], poses difficulties. From the 1703 description given it appears that the case being referred to in 1704 one in which two network addresses return server_owners with the 1705 same major_id and different minor_id values, making the network 1706 addresses server-trunkable without being session trunkable. 1708 A better approach would be to refer to "server-trunking" as used 1709 elsewhere in this document and use the replacement for the 1710 existing Section 18.35 to identify clientid trunking as the means 1711 to adapt to network addresses which are server-trunkable without 1712 being session-trunkable and session trunking as the means to adapt 1713 to network addresses which are session-trunkable. 1715 o The existing Section 11.7.2 needs to be better connected to 1716 trunking discovery. By calling these "transparent" transitions, 1717 it obscures the fact that some (or all) of the "transitions" it is 1718 discussing are not in fact transitions between servers or file 1719 systems but merely changes the set of communication paths in use. 1721 o The existing Section 11.7.2.1, needs to address more clearly the 1722 case of server-trunkable addresses which are not session- 1723 trunkable. As it is, it mentions the related concept of 1724 clustering, but only deals explicitly with the case in which two 1725 distinct servers share access to one or more file systems and does 1726 not mention the case in which the network addresses can be used to 1727 access a shared stateid space without being session-trunkable. 1729 o The existing Section 11.7.2.2, while correct, needs to be part of 1730 a general re-organization since the characteristics it lists as 1731 necessary for a transparent transition will be of use in other 1732 contexts, particularly as they apply to Transparent State 1733 Migration as well. It make sense to move these to a new sub- 1734 section within the equivalent of the Existing Section 11.7. 1736 o The existing Section 11.7.7, needs the a major rework to deal with 1737 its basic assumption, that existing state can only be made 1738 available on the destination server if the source and destination 1739 co-operate in state management and maintain a common client id 1740 space. It is not clear how this can be done, other than for 1741 servers working together so as to provide clientid trunking, a 1742 case that is probably considered as a "transparent transition". 1743 The section needs to modified to allow something along the lines 1744 of NFSv4.0-style Transparent State Migration with the details 1745 provided by a later section (see below). 1747 A related issues concerns the sentence, "In the case of migration, 1748 the servers involved in the migration of a file system SHOULD 1749 transfer all server state from the original to the new server. It 1750 is unclear why this is a "SHOULD" as the rest of the paragraph 1751 essentially tells the client that it needs to be prepared for the 1752 server not to do this. The equivalent is a "should" in [RFC7931], 1753 and there is no reason to add to confusion by making a "SHOULD" in 1754 NFSv4.1. also, there is no mention of the need to provide a fs- 1755 specific grace period in the cases in which Transparent State 1756 Migration is not made available. 1758 o Adding a new section (at level of the existing Section 11.7.7) 1759 about state transfer during migration. Although the phrase 1760 "Transparent State Migration" is well established in the context 1761 of NFSv4.0, the word "transparent" could cause confusion given the 1762 existing use of the phrase "transparent transitions". A possible 1763 title for the new section is "State Transfer during Migration" 1765 The new section would present the NFSv4.1-equivalent of Transparent 1766 State Migration as described in [RFC7931]. This would address the 1767 issues presented in Section 5.1 along the lines suggested in Sections 1768 5.2, 5.3, and 5.4. 1770 6. Security Considerations 1772 With regard to NFSv4.0, the Security Considerations section of 1773 [RFC7530] as modified by that of [RFC7931] takes proper care of 1774 migration-related issues. No change is needed. 1776 With regard to NFSv4.1, the Security Considerations section of 1777 [RFC5661] takes proper care of migration-related issues. No change 1778 is needed. 1780 7. IANA Considerations 1782 This document does not require actions by IANA. 1784 8. Normative References 1786 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1787 Requirement Levels", BCP 14, RFC 2119, 1788 DOI 10.17487/RFC2119, March 1997, 1789 . 1791 [RFC5661] Shepler, S., Ed., Eisler, M., Ed., and D. Noveck, Ed., 1792 "Network File System (NFS) Version 4 Minor Version 1 1793 Protocol", RFC 5661, DOI 10.17487/RFC5661, January 2010, 1794 . 1796 [RFC7530] Haynes, T., Ed. and D. Noveck, Ed., "Network File System 1797 (NFS) Version 4 Protocol", RFC 7530, DOI 10.17487/RFC7530, 1798 March 2015, . 1800 [RFC7931] Noveck, D., Ed., Shivam, P., Lever, C., and B. Baker, 1801 "NFSv4.0 Migration: Specification Update", RFC 7931, 1802 DOI 10.17487/RFC7931, July 2016, 1803 . 1805 Appendix A. Acknowledgements 1807 The editor and authors of this document gratefully acknowledge the 1808 contributions of Trond Myklebust of Primary Data, Robert Thurlow of 1809 Oracle, and Andy Adamson of NetApp. We also thank Tom Haynes of 1810 Primary Data and Spencer Shepler of Microsoft for their guidance and 1811 suggestions. 1813 Special thanks go to members of the Oracle Solaris NFS team, 1814 especially Rick Mesta and James Wahlig, for their work implementing 1815 an NFSv4.0 migration prototype and identifying many of the issues 1816 documented here. Also, the work of Xuan Qi of Oracle with NFSv4.1 1817 client and server prototypes was helpful. 1819 Appendix B. History of this Document 1821 The contents of successive versions of this document have changed 1822 because new issues have been discovered, because there here have 1823 changes in our understanding of how these features should interact, 1824 and because some of the issues have been adequately addressed with 1825 regard to certain protocol versions. 1827 As a result, it may be helpful to understand the history of these 1828 issues, which is complicated because multiple NFSv4 protocols have 1829 been involved. 1831 This history can be summarized as follows 1833 o Initially, the focus was on the difficulties seen in NFSv4.0 1834 implementations of Transparent State Migration, and on identifying 1835 possible corrections to [RFC7530] that might address these issues. 1837 At this point, treatment of NFSv4.1 was minimal. 1839 o As examination of the issues continued, it became clear that the 1840 use of the non-uniform client string model was a critical element 1841 of the problem and further work proceeded on that basis. 1843 During the period, treatment of NFSv4.1 was expanded but the fact 1844 that NFSv4.1 had existing facilities for trunking detection was 1845 taken as an indication that the problems would not be difficult to 1846 address.. 1848 o As work proceeded on a standards-track document addressing the 1849 NFSv4.0 issues, material that proposed changes to address the 1850 issues became less relevant, since the effective vehicle for 1851 addressing these issues became the standards-track document. 1853 During this period, and subsequently, treatment of NFSv4.1 1854 remained essentially unchanged. 1856 o With the publication of [RFC7931], material regarding fixes for 1857 the NSV4.0 became vestigial but the material was retained for a 1858 while together with a shift from proposing those changes to 1859 reporting that they had been made. 1861 o Later, in response to experiences testing existing NFSv4.1 1862 implementations of migration, the focus of the document shifted 1863 decisively to NFSv4.1. As part of the analysis of migration 1864 within NFSv4.1, it was realized that issues related to the 1865 appearance of multiple addresses were fundamental to clearly 1866 describing how migration would work and that changes in the set of 1867 such addresses might or might not involve migration. 1869 At this point, discussion of NFSV4.0 issues was further limited. 1870 The issues seen were noted but the discussion of the resolution 1871 was limited to explaining that they had been addressed by the 1872 publication of [RFC7931]. 1874 o Finally, based on the results of work to provide NFSv4 with 1875 trunking discovery facilities, a decision was made that this work 1876 was most appropriately dealt with together with migration, for 1877 reasons noted previously. 1879 Since the trunking discovery facilities apply to all NFSv4 minor 1880 versions, work was needed to define those for NFSv4.0as well, 1881 together with the necessary interactions with migration. 1883 Authors' Addresses 1885 David Noveck (editor) 1886 NetApp 1887 26 Locust Avenue 1888 Lexington, MA 02421 1889 US 1891 Phone: +1 781 572 8038 1892 Email: davenoveck@gmail.com 1894 Piyush Shivam 1895 Oracle Corporation 1896 5300 Riata Park Ct. 1897 Austin, TX 78727 1898 US 1900 Phone: +1 512 401 1019 1901 Email: piyush.shivam@oracle.com 1902 Charles Lever 1903 Oracle Corporation 1904 1015 Granger Avenue 1905 Ann Arbor, MI 48104 1906 US 1908 Phone: +1 248 614 5091 1909 Email: chuck.lever@oracle.com 1911 Bill Baker 1912 Oracle Corporation 1913 5300 Riata Park Ct. 1914 Austin, TX 78727 1915 US 1917 Phone: +1 512 401 1081 1918 Email: bill.baker@oracle.com