idnits 2.17.1 draft-ietf-nfsv4-rfc5667bis-06.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 seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (February 24, 2017) is 2616 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-11) exists of draft-ietf-nfsv4-rfc5666bis-10 == Outdated reference: A later version (-08) exists of draft-ietf-nfsv4-rpcrdma-bidirection-07 ** Obsolete normative reference: RFC 5661 (Obsoleted by RFC 8881) == Outdated reference: A later version (-11) exists of draft-ietf-nfsv4-versioning-09 -- Obsolete informational reference (is this intentional?): RFC 5666 (Obsoleted by RFC 8166) -- Obsolete informational reference (is this intentional?): RFC 5667 (Obsoleted by RFC 8267) Summary: 1 error (**), 0 flaws (~~), 5 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network File System Version 4 C. Lever, Ed. 3 Internet-Draft Oracle 4 Obsoletes: 5667 (if approved) February 24, 2017 5 Intended status: Standards Track 6 Expires: August 28, 2017 8 Network File System (NFS) Upper Layer Binding To RPC-Over-RDMA Version 9 One 10 draft-ietf-nfsv4-rfc5667bis-06 12 Abstract 14 This document specifies Upper Layer Bindings of Network File System 15 (NFS) protocol versions to RPC-over-RDMA Version One. Upper Layer 16 Bindings are required in order to enable RPC-based protocols such as 17 NFS to use Direct Data Placement on RPC-over-RDMA Version One. This 18 document obsoletes RFC 5667. 20 Requirements Language 22 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 23 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 24 document are to be interpreted as described in [RFC2119]. 26 Status of This Memo 28 This Internet-Draft is submitted in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF). Note that other groups may also distribute 33 working documents as Internet-Drafts. The list of current Internet- 34 Drafts is at http://datatracker.ietf.org/drafts/current/. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 This Internet-Draft will expire on August 28, 2017. 43 Copyright Notice 45 Copyright (c) 2017 IETF Trust and the persons identified as the 46 document authors. All rights reserved. 48 This document is subject to BCP 78 and the IETF Trust's Legal 49 Provisions Relating to IETF Documents 50 (http://trustee.ietf.org/license-info) in effect on the date of 51 publication of this document. Please review these documents 52 carefully, as they describe your rights and restrictions with respect 53 to this document. Code Components extracted from this document must 54 include Simplified BSD License text as described in Section 4.e of 55 the Trust Legal Provisions and are provided without warranty as 56 described in the Simplified BSD License. 58 This document may contain material from IETF Documents or IETF 59 Contributions published or made publicly available before November 60 10, 2008. The person(s) controlling the copyright in some of this 61 material may not have granted the IETF Trust the right to allow 62 modifications of such material outside the IETF Standards Process. 63 Without obtaining an adequate license from the person(s) controlling 64 the copyright in such materials, this document may not be modified 65 outside the IETF Standards Process, and derivative works of it may 66 not be created outside the IETF Standards Process, except to format 67 it for publication as an RFC or to translate it into languages other 68 than English. 70 Table of Contents 72 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 73 2. Reply Size Estimation . . . . . . . . . . . . . . . . . . . . 3 74 2.1. Short Reply Chunk Retry . . . . . . . . . . . . . . . . . 4 75 3. Upper Layer Binding for NFS Versions 2 and 3 . . . . . . . . 5 76 3.1. Reply Size Estimation . . . . . . . . . . . . . . . . . . 5 77 3.2. RPC Binding Considerations . . . . . . . . . . . . . . . 5 78 4. Upper Layer Bindings for NFS Version 2 and 3 Auxiliary 79 Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . 6 80 4.1. MOUNT, NLM, and NSM Protocols . . . . . . . . . . . . . . 6 81 4.2. NFSACL Protocol . . . . . . . . . . . . . . . . . . . . . 7 82 5. Upper Layer Binding For NFS Version 4 . . . . . . . . . . . . 7 83 5.1. DDP-Eligibility . . . . . . . . . . . . . . . . . . . . . 7 84 5.2. Reply Size Estimation . . . . . . . . . . . . . . . . . . 8 85 5.3. RPC Binding Considerations . . . . . . . . . . . . . . . 9 86 5.4. NFS COMPOUND Requests . . . . . . . . . . . . . . . . . . 10 87 5.5. NFS Callback Requests . . . . . . . . . . . . . . . . . . 11 88 5.6. Session-Related Considerations . . . . . . . . . . . . . 12 89 5.7. Transport Considerations . . . . . . . . . . . . . . . . 13 90 6. Extending NFS Upper Layer Bindings . . . . . . . . . . . . . 14 91 7. Security Considerations . . . . . . . . . . . . . . . . . . . 14 92 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15 93 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 15 94 9.1. Normative References . . . . . . . . . . . . . . . . . . 15 95 9.2. Informative References . . . . . . . . . . . . . . . . . 16 97 Appendix A. Changes Since RFC 5667 . . . . . . . . . . . . . . . 17 98 Appendix B. Acknowledgments . . . . . . . . . . . . . . . . . . 18 99 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 19 101 1. Introduction 103 An RPC-over-RDMA Version One transport may employ direct data 104 placement to convey certain data payloads associated with RPC 105 transactions [I-D.ietf-nfsv4-rfc5666bis]. To enable successful 106 interoperation, implementations of RPC Programs running on RPC-over- 107 RDMA must agree as to which XDR data items in what particular RPC 108 procedures are eligible for direct data placement (DDP). This 109 agreement is specified in an Upper Layer Binding. 111 This document contains material required of Upper Layer Bindings, as 112 specified in [I-D.ietf-nfsv4-rfc5666bis], for the following NFS 113 protocol versions: 115 o NFS Version 2 [RFC1094] 117 o NFS Version 3 [RFC1813] 119 o NFS Version 4.0 [RFC7530] 121 o NFS Version 4.1 [RFC5661] 123 o NFS Version 4.2 [RFC7862] 125 This document assumes the reader is already familiar with concepts 126 and terminology defined in [I-D.ietf-nfsv4-rfc5666bis] and the 127 documents it references. 129 2. Reply Size Estimation 131 On an RPC-over-RDMA Version One transport, during the construction of 132 each RPC Call message, a requester is responsible for allocating 133 appropriate resources for receiving the matching Reply message. 135 An overrun of these resources can result in corruption of the Reply 136 message or termination of the transport connection. Therefore 137 reliable reply size estimation is necessary to ensure successful 138 interoperation. This is particularly critical, for example, when 139 allocating a Reply chunk. 141 In most cases, the NFS protocol's XDR definition provides enough 142 information to enable an NFS client to predict the maximum size of 143 the expected Reply message. If there are variable-size data items in 144 the result, the maximum size of the RPC Reply message can be 145 estimated as follows: 147 o The client requests only a specific portion of an object (for 148 example, using the "count" and "offset" fields in an NFS READ). 150 o The client has already cached the size of the whole object it is 151 about to request (say, via a previous NFS GETATTR request). 153 o The client and server have negotiated a maximum size for all calls 154 and responses (using a CREATE_SESSION operation, for instance). 156 2.1. Short Reply Chunk Retry 158 In a few cases, either the size of one or more returned data items or 159 the number of returned data items cannot be known in advance of 160 forming an RPC Call. 162 A requester uses a Reply chunk to handle an RPC transaction where the 163 expected RPC Reply message might be larger than the requester's 164 inline threshold. If an actual RPC Reply message does not fit in a 165 client-provided Reply chunk, the NFS server responds with an 166 RDMA_ERROR message with the rdma_err field set to ERR_CHUNK, or it 167 could even break the transport connection. 169 In response, an NFS client can choose to: 171 o Terminate the RPC transaction with an error, or 173 o Allocate a larger Reply chunk and send the same request as a new 174 RPC transaction (to avoid hitting in a Duplicate Reply Cache). 175 The NFS client should avoid retrying the request indefinitely 176 because a responder may return ERR_CHUNK for a variety of reasons. 178 The latter choice is considered heroic recovery, and is only a real 179 choice for the few operations where it is not possible for an NFS 180 client to predict the size of the Reply message in advance. 182 Subsequent sections of this document discuss exactly which operations 183 might have ultimate difficulty with Reply size estimation. These 184 operations are eligible for "short Reply chunk retry." Unless 185 explicitly mentioned as applicable, short Reply chunk retry should 186 not be used. 188 3. Upper Layer Binding for NFS Versions 2 and 3 190 The Upper Layer Binding specification in this section applies to NFS 191 Version 2 [RFC1094] and NFS Version 3 [RFC1813]. For brevity, in 192 this document a "Legacy NFS client" refers to an NFS client using the 193 NFS version 2 or NFS version 3 RPC Programs (100003) to communicate 194 with an NFS server. Likewise, a "Legacy NFS server" is an NFS server 195 communicating with clients using NFS version 2 or NFS version 3. 197 The following XDR data items in NFS versions 2 and 3 are DDP- 198 eligible: 200 o The opaque file data argument in the NFS WRITE procedure 202 o The pathname argument in the NFS SYMLINK procedure 204 o The opaque file data result in the NFS READ procedure 206 o The pathname result in the NFS READLINK procedure 208 All other argument or result data items in NFS versions 2 and 3 are 209 not DDP-eligible. 211 A Legacy NFS client MUST NOT send a reduced Payload stream in a Long 212 Call. A Legacy NFS client MUST NOT enable a Legacy NFS server to 213 send a reduced Payload stream in a Long Reply. 215 A Legacy server's response to a DDP-eligibility violation does not 216 give an indication to Legacy clients of whether the server has 217 processed the arguments of the RPC Call, or whether the server has 218 accessed or modified client memory associated with that RPC. 220 3.1. Reply Size Estimation 222 A Legacy NFS client determines the maximum reply size for each 223 operation using the criteria outlined in Section 2. There are no 224 operations in NFS version 2 or 3 that benefit from short Reply chunk 225 retry. 227 3.2. RPC Binding Considerations 229 Legacy NFS servers traditionally listen for clients on UDP and TCP 230 port 2049. Additionally, they register these ports with a local 231 portmapper [RFC1833] service. 233 A Legacy NFS server supporting RPC-over-RDMA Version One on such a 234 network and registering itself with the RPC portmapper MAY choose an 235 arbitrary port, or MAY use the alternative well-known port number for 236 its RPC-over-RDMA service (see Section 8). The chosen port MAY be 237 registered with the RPC portmapper under the netids assigned in 238 [I-D.ietf-nfsv4-rfc5666bis]. 240 4. Upper Layer Bindings for NFS Version 2 and 3 Auxiliary Protocols 242 NFS versions 2 and 3 are typically deployed with several other 243 protocols, sometimes referred to as "NFS auxiliary protocols." These 244 are distinct RPC Programs that define procedures which are not part 245 of the NFS version 2 or version 3 RPC Programs. The Upper Layer 246 Bindings in this section apply to: 248 o The MOUNT and NLM protocols, introduced in an appendix of 249 [RFC1813] 251 o The NSM protocol, described in Chapter 11 of [NSM] 253 o The NFSACL protocol, which does not have a public definition. 254 NFSACL is treated in this document as a de facto standard, as 255 there are several interoperating implementations. 257 RPC-over-RDMA Version One considers these RPC Programs as separate 258 Upper Layer Protocols [I-D.ietf-nfsv4-rfc5666bis]. Therefore a 259 separate Upper Layer Binding, provided here, is required for each of 260 these. 262 4.1. MOUNT, NLM, and NSM Protocols 264 Typically MOUNT, NLM, and NSM are conveyed via TCP, even in 265 deployments where the NFS RPC Program operates on RPC-over-RDMA 266 Version One. When a Legacy server supports these RPC Programs on 267 RPC-over-RDMA Version One, it advertises the port address via the 268 usual rpcbind service [RFC1833]. 270 No operation in these protocols conveys a significant data payload, 271 and the size of RPC messages in these protocols is uniformly small. 272 Therefore, no XDR data items in these protocols are DDP-eligible. 274 The largest variable-length XDR data item is an xdr_netobj. In most 275 implementations this data item is never larger than 1024 bytes, 276 making reliable reply size estimation straightforward using the 277 criteria outlined in Section 2. There are no operations in these 278 protocols that benefit from short Reply chunk retry. 280 4.2. NFSACL Protocol 282 Legacy clients and servers that support the NFSACL RPC Program 283 typically convey NFSACL procedures on the same connection as NFS RPC 284 Programs. This obviates the need for separate rpcbind queries to 285 discover server support for this RPC Program. 287 ACLs are typically small, but even large ACLs must be encoded and 288 decoded to some degree. Thus no data item in this Upper Layer 289 Protocol is DDP-eligible. 291 For procedures whose replies do not include an ACL object, the size 292 of a reply is determined directly from the NFSACL RPC Program's XDR 293 definition. 295 There is no protocol-specified size limit for NFS version 3 ACLs, and 296 there is no mechanism in either the NFSACL or NFS RPC Programs for a 297 Legacy client to ascertain the largest ACL a Legacy server can 298 return. Legacy client implementations should choose a maximum size 299 for ACLs based on their own internal limits. 301 Because an NFSACL client cannot know in advance how large a returned 302 ACL will be, it can use short Reply chunk retry when an NFSACL GETACL 303 operation encounters a transport error. 305 5. Upper Layer Binding For NFS Version 4 307 The Upper Layer Binding specification in this section applies to RPC 308 Programs defined in NFS Version 4.0 [RFC7530], NFS Version 4.1 309 [RFC5661], and NFS Version 4.2 [RFC7862]. 311 5.1. DDP-Eligibility 313 Only the following XDR data items in the COMPOUND procedure of all 314 NFS version 4 minor versions are DDP-eligible: 316 o The opaque data field in the WRITE4args structure 318 o The linkdata field of the NF4LNK arm in the createtype4 union 320 o The opaque data field in the READ4resok structure 322 o The linkdata field in the READLINK4resok structure 324 o In minor version 2 and newer, the rpc_data field of the 325 read_plus_content union (further restrictions on the use of this 326 data item follow below). 328 5.1.1. READ_PLUS Replies 330 The NFS version 4.2 READ_PLUS operation returns a complex data type 331 [RFC7862]. The rpr_contents field in the result of this operation is 332 an array of read_plus_content unions, one arm of which contains an 333 opaque byte stream (d_data). 335 The size of d_data is limited to the value of the rpa_count field, 336 but the protocol does not bound the number of elements which can be 337 returned in the rpr_contents array. In order to make the size of 338 READ_PLUS replies predictable by NFS version 4.2 clients, the 339 following restrictions are placed on the use of the READ_PLUS 340 operation on an RPC-over-RDMA Version One transport: 342 o An NFS version 4.2 client MUST NOT provide more than one Write 343 chunk for any READ_PLUS operation. When providing a Write chunk 344 for a READ_PLUS operation, an NFS version 4.2 client MUST provide 345 a Write chunk that is either empty (which forces all result data 346 items for this operation to be returned inline) or large enough to 347 receive rpa_count bytes in a single element of the rpr_contents 348 array. 350 o If the Write chunk provided for a READ_PLUS operation by an NFS 351 version 4.2 client is not empty, an NFS version 4.2 server MUST 352 use that chunk for the first element of the rpr_contents array 353 that has an rpc_data arm. 355 o An NFS version 4.2 server MUST NOT return more than two elements 356 in the rpr_contents array of any READ_PLUS operation. It returns 357 as much of the requested byte range as it can fit within these two 358 elements. If the NFS version 4.2 server has not asserted rpr_eof 359 in the reply, the NFS version 4.2 client SHOULD send additional 360 READ_PLUS requests for any remaining bytes. 362 5.2. Reply Size Estimation 364 Within NFS version 4, there are certain variable-length result data 365 items whose maximum size cannot be estimated by clients reliably 366 because there is no protocol-specified size limit on these arrays. 367 These include: 369 o The attrlist4 field 371 o Fields containing ACLs such as fattr4_acl, fattr4_dacl, 372 fattr4_sacl 374 o Fields in the fs_locations4 and fs_locations_info4 data structures 375 o Fields opaque to the NFS version 4 protocol which pertain to pNFS 376 layout metadata, such as loc_body, loh_body, da_addr_body, 377 lou_body, lrf_body, fattr_layout_types and fs_layout_types, 379 5.2.1. Reply Size Estimation for Minor Version 0 381 The NFS version 4.0 protocol itself does not impose any bound on the 382 size of NFS calls or responses. 384 Some of the data items enumerated in Section 5.2 (in particular, the 385 items related to ACLs and fs_locations) make it difficult to predict 386 the maximum size of NFS version 4.0 replies that interrogate 387 variable-length fattr4 attributes. As discussed in Section 2, client 388 implementations can rely on their own internal architectural limits 389 to constrain the reply size, but such limits are not always 390 guaranteed to be reliable. 392 When an especially large fattr4 result is expected, a Reply chunk 393 might be required. An NFS version 4.0 client can use short Reply 394 chunk retry when an NFS COMPOUND containing a GETATTR operation 395 encounters a transport error. 397 The use of NFS COMPOUND operations raises the possibility of requests 398 that combine a non-idempotent operation (e.g. WRITE) with a GETATTR 399 operation that requests one or more variable-length results. This 400 combination should be avoided by ensuring that any GETATTR operation 401 that requests a result of unpredictable length is sent in an NFS 402 COMPOUND by itself. 404 5.2.2. Reply Size Estimation for Minor Version 1 and Newer 406 In NFS version 4.1 and newer minor versions, the csa_fore_chan_attrs 407 argument of the CREATE_SESSION operation contains a 408 ca_maxresponsesize field. The value in this field can be taken as 409 the absolute maximum size of replies generated by an NFS version 4.1 410 server. 412 This value can be used in cases where it is not possible to estimate 413 a reply size upper bound precisely. In practice, objects such as 414 ACLs, named attributes, layout bodies, and security labels are much 415 smaller than this maximum. 417 5.3. RPC Binding Considerations 419 NFS version 4 servers are required to listen on TCP port 2049, and 420 they are not required to register with an rpcbind service [RFC7530]. 422 Therefore, an NFS version 4 server supporting RPC-over-RDMA Version 423 One MUST use the alternative well-known port number for its RPC-over- 424 RDMA service (see Section 8). Clients SHOULD connect to this well- 425 known port without consulting the RPC portmapper (as for NFS version 426 4 on TCP transports). 428 5.4. NFS COMPOUND Requests 430 5.4.1. Long Calls and Replies 432 Each NFS version 4 COMPOUND procedure contains an array of operations 433 which may be larger than a connection's inline thresholds, even after 434 reduction of DDP-elibible payloads. Therefore, an NFS version 4 435 client MAY send a reduced Payload stream in a Long Call. An NFS 436 version 4 client MAY enable an NFS version 4 server to send a reduced 437 Payload stream in a Long Reply. 439 5.4.2. Multiple DDP-eligible Data Items 441 The NFS version 4 COMPOUND procedure allows the transmission of more 442 than one DDP-eligible data item per Call and Reply message. An NFS 443 version 4 client provides XDR Position values in each Read chunk to 444 disambiguate which chunk is associated with which argument data item. 446 However NFS version 4 server and client implementations must agree in 447 advance on how to pair Write chunks with returned result data items. 448 The mechanism specified in Section 4.3.2 of 449 [I-D.ietf-nfsv4-rfc5666bis]) is applied here, with additional 450 restrictions that appear below. 452 In the following list, an "NFS Read" operation refers to any NFS 453 Version 4 operation which has a DDP-eligible result data item (i.e., 454 either a READ, READ_PLUS, or READLINK operation). 456 o If an NFS version 4 client wishes all DDP-eligible items in an NFS 457 reply to be conveyed inline, it leaves the Write list empty. 459 o The first chunk in the Write list MUST be used by the first READ 460 operation in an NFS version 4 COMPOUND procedure. The next Write 461 chunk is used by the next READ operation, and so on. 463 o If an NFS version 4 client has provided a matching non-empty Write 464 chunk, then the corresponding READ operation MUST return its DDP- 465 eligible data item using that chunk. 467 o If an NFS version 4 client has provided an empty matching Write 468 chunk, then the corresponding READ operation MUST return all of 469 its result data items inline. 471 o If a READ operation returns a union arm which does not contain a 472 DDP-eligible result, and the NFS version 4 client has provided a 473 matching non-empty Write chunk, an NFS version 4 server MUST 474 return an empty Write chunk in that Write list position. 476 o If there are more READ operations than Write chunks, then 477 remaining NFS Read operations in an NFS version 4 COMPOUND that 478 have no matching Write chunk MUST return their results inline. 480 5.4.3. NFS Version 4 COMPOUND Example 482 The following example shows a Write list with three Write chunks, A, 483 B, and C. The NFS version 4 server consumes the provided Write 484 chunks by writing the results of the designated operations in the 485 compound request (READ and READLINK) back to each chunk. 487 Write list: 489 A --> B --> C 491 NFS version 4 COMPOUND request: 493 PUTFH LOOKUP READ PUTFH LOOKUP READLINK PUTFH LOOKUP READ 494 | | | 495 v v v 496 A B C 498 If the NFS version 4 client does not want to have the READLINK result 499 returned via RDMA, it provides an empty Write chunk for buffer B to 500 indicate that the READLINK result must be returned inline. 502 5.5. NFS Callback Requests 504 The NFS version 4 family of protocols support server-initiated 505 callbacks to notify NFS version 4 clients of events such as recalled 506 delegations. 508 5.5.1. NFS Version 4.0 Callback 510 NFS version 4.0 implementations typically employ a separate TCP 511 connection to handle callback operations, even when the forward 512 channel uses an RPC-over-RDMA Version One transport. 514 No operation in the NFS version 4.0 callback RPC Program conveys a 515 significant data payload. Therefore, no XDR data items in this RPC 516 Program is DDP-eligible. 518 A CB_RECALL reply is small and fixed in size. The CB_GETATTR reply 519 contains a variable-length fattr4 data item. See Section 5.2.1 for a 520 discussion of reply size prediction for this data item. 522 An NFS version 4.0 client advertises netids and ad hoc port addresses 523 for contacting its NFS version 4.0 callback service using the 524 SETCLIENTID operation. 526 5.5.2. NFS Version 4.1 Callback 528 In NFS version 4.1 and newer minor versions, callback operations may 529 appear on the same connection as is used for NFS version 4 forward 530 channel client requests. NFS version 4 clients and servers MUST use 531 the mechanism described in [I-D.ietf-nfsv4-rpcrdma-bidirection] when 532 backchannel operations are conveyed on RPC-over-RDMA Version One 533 transports. 535 The csa_back_chan_attrs argument of the CREATE_SESSION operation 536 contains a ca_maxresponsesize field. The value in this field can be 537 taken as the absolute maximum size of backchannel replies generated 538 by a replying NFS version 4 client. 540 There are no DDP-eligible data items in callback procedures defined 541 in NFS version 4.1 or NFS version 4.2. However, some callback 542 operations, such as messages that convey device ID information, can 543 be large, in which case a Long Call or Reply might be required. 545 When an NFS version 4.1 client can support Long Calls in its 546 backchannel, it reports a backchannel ca_maxrequestsize that is 547 larger than the connection's inline thresholds. Otherwise an NFS 548 version 4 server MUST use only Short messages to convey backchannel 549 operations. 551 5.6. Session-Related Considerations 553 The presence of an NFS session (defined in [RFC5661]) has no effect 554 on the operation of RPC-over-RDMA Version One. None of the 555 operations introduced to support NFS sessions (e.g. the SEQUENCE 556 operation) contain DDP-eligible data items. There is no need to 557 match the number of session slots with the number of available RPC- 558 over-RDMA credits. 560 However, there are a few new cases where an RPC transaction can fail. 561 For example, a requester might receive, in response to an RPC 562 request, an RDMA_ERROR message with an rdma_err value of ERR_CHUNK, 563 or an RDMA_MSG containing an RPC_GARBAGEARGS reply. These situations 564 are no different from existing RPC errors which an NFS session 565 implementation is already prepared to handle for other transports. 567 And as with other transports during such a failure, there might be no 568 SEQUENCE result available to the requester to distinguish whether 569 failure occurred before or after the requested operations were 570 executed on the responder. 572 When a transport error occurs (e.g. RDMA_ERROR), the requester 573 proceeds as usual to match the incoming XID value to a waiting RPC 574 Call. The RPC transaction is terminated, and the result status is 575 reported to the Upper Layer Protocol. The requester's session 576 implementation then determines the session ID and slot for the failed 577 request, and performs slot recovery to make that slot usable again. 578 If this is not done, that slot could be rendered permanently 579 unavailable. 581 5.7. Transport Considerations 583 5.7.1. Congestion Avoidance 585 Section 3.1 of [RFC7530] states: 587 Where an NFSv4 implementation supports operation over the IP 588 network protocol, the supported transport layer between NFS and IP 589 MUST be an IETF standardized transport protocol that is specified 590 to avoid network congestion; such transports include TCP and the 591 Stream Control Transmission Protocol (SCTP). 593 Section 2.9.1 of [RFC5661] further states: 595 Even if NFSv4.1 is used over a non-IP network protocol, it is 596 RECOMMENDED that the transport support congestion control. 598 It is permissible for a connectionless transport to be used under 599 NFSv4.1; however, reliable and in-order delivery of data combined 600 with congestion control by the connectionless transport is 601 REQUIRED. As a consequence, UDP by itself MUST NOT be used as an 602 NFSv4.1 transport. 604 RPC-over-RDMA Version One is constructed on a platform of RDMA 605 Reliable Connections [I-D.ietf-nfsv4-rfc5666bis] [RFC5041]. RDMA 606 Reliable Connections are reliable, connection-oriented transports 607 that guarantee in-order delivery, meeting all above requirements for 608 NFS version 4 transports. 610 5.7.2. Retransmission and Keep-alive 612 NFS version 4 client implementations often rely on a transport-layer 613 keep-alive mechanism to detect when an NFS version 4 server has 614 become unresponsive. When an NFS server is no longer responsive, 615 client-side keep-alive terminates the connection, which in turn 616 triggers reconnection and RPC retransmission. 618 Some RDMA transports (such as Reliable Connections on InfiniBand) 619 have no keep-alive mechanism. Without a disconnect or new RPC 620 traffic, such connections can remain alive long after an NFS server 621 has become unresponsive. Once an NFS client has consumed all 622 available RPC-over-RDMA credits on that transport connection, it will 623 forever await a reply before sending another RPC request. 625 NFS version 4 clients SHOULD reserve one RPC-over-RDMA credit to use 626 for periodic server or connection health assessment. This credit can 627 be used to drive an RPC request on an otherwise idle connection, 628 triggering either a quick affirmative server response or immediate 629 connection termination. 631 In addition to network partition and request loss scenarios, RPC- 632 over-RDMA transport connections can be terminated when a Transport 633 header is malformed, Reply messages are larger than receive 634 resources, or when too many RPC-over-RDMA messages are sent at once. 635 In such cases: 637 o If there is a transport error indicated (ie, RDMA_ERROR) before 638 the disconnect or instead of a disconnect, the requester MUST 639 respond to that error as prescribed by the specification of the 640 RPC transport. Then the NFS version 4 rules for handling 641 retransmission apply. 643 o If there is a transport disconnect and the responder has provided 644 no other response for a request, then only the NFS version 4 rules 645 for handling retransmission apply. 647 6. Extending NFS Upper Layer Bindings 649 RPC Programs such as NFS are required to have an Upper Layer Binding 650 specification to interoperate on RPC-over-RDMA Version One transports 651 [I-D.ietf-nfsv4-rfc5666bis]. Via standards action, the Upper Layer 652 Binding specified in this document can be extended to cover versions 653 of the NFS version 4 protocol specified after NFS version 4 minor 654 version 2, or separately published extensions to an existing NFS 655 version 4 minor version, as described in [I-D.ietf-nfsv4-versioning]. 657 7. Security Considerations 659 RPC-over-RDMA Version One supports all RPC security models, including 660 RPCSEC_GSS security and transport-level security [RFC2203]. The 661 choice of what Direct Data Placement mechanism to convey RPC argument 662 and results does not affect this, since it changes only the method of 663 data transfer. Specifically, the requirements of 664 [I-D.ietf-nfsv4-rfc5666bis] ensure that this choice does not 665 introduce new vulnerabilities. 667 Because this document defines only the binding of the NFS protocols 668 atop [I-D.ietf-nfsv4-rfc5666bis], all relevant security 669 considerations are therefore to be described at that layer. 671 8. IANA Considerations 673 The use of direct data placement in NFS introduces a need for an 674 additional port number assignment for networks that share traditional 675 UDP and TCP port spaces with RDMA services. The iWARP protocol is 676 such an example [RFC5041] [RFC5040]. 678 For this purpose, a set of transport protocol port number assignments 679 is specified by this document. IANA has assigned the following ports 680 for NFS/RDMA in the IANA port registry, according to the guidelines 681 described in [RFC6335]. 683 nfsrdma 20049/tcp Network File System (NFS) over RDMA 684 nfsrdma 20049/udp Network File System (NFS) over RDMA 685 nfsrdma 20049/sctp Network File System (NFS) over RDMA 687 This document should be listed as the reference for the nfsrdma port 688 assignments. This document does not alter these assignments. 690 9. References 692 9.1. Normative References 694 [I-D.ietf-nfsv4-rfc5666bis] 695 Lever, C., Simpson, W., and T. Talpey, "Remote Direct 696 Memory Access Transport for Remote Procedure Call, Version 697 One", draft-ietf-nfsv4-rfc5666bis-10 (work in progress), 698 February 2017. 700 [I-D.ietf-nfsv4-rpcrdma-bidirection] 701 Lever, C., "Bi-directional Remote Procedure Call On RPC- 702 over-RDMA Transports", draft-ietf-nfsv4-rpcrdma- 703 bidirection-07 (work in progress), February 2017. 705 [RFC1833] Srinivasan, R., "Binding Protocols for ONC RPC Version 2", 706 RFC 1833, DOI 10.17487/RFC1833, August 1995, 707 . 709 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 710 Requirement Levels", BCP 14, RFC 2119, 711 DOI 10.17487/RFC2119, March 1997, 712 . 714 [RFC2203] Eisler, M., Chiu, A., and L. Ling, "RPCSEC_GSS Protocol 715 Specification", RFC 2203, DOI 10.17487/RFC2203, September 716 1997, . 718 [RFC5661] Shepler, S., Ed., Eisler, M., Ed., and D. Noveck, Ed., 719 "Network File System (NFS) Version 4 Minor Version 1 720 Protocol", RFC 5661, DOI 10.17487/RFC5661, January 2010, 721 . 723 [RFC6335] Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S. 724 Cheshire, "Internet Assigned Numbers Authority (IANA) 725 Procedures for the Management of the Service Name and 726 Transport Protocol Port Number Registry", BCP 165, 727 RFC 6335, DOI 10.17487/RFC6335, August 2011, 728 . 730 [RFC7530] Haynes, T., Ed. and D. Noveck, Ed., "Network File System 731 (NFS) Version 4 Protocol", RFC 7530, DOI 10.17487/RFC7530, 732 March 2015, . 734 [RFC7862] Haynes, T., "Network File System (NFS) Version 4 Minor 735 Version 2 Protocol", RFC 7862, DOI 10.17487/RFC7862, 736 November 2016, . 738 9.2. Informative References 740 [I-D.ietf-nfsv4-versioning] 741 Noveck, D., "Rules for NFSv4 Extensions and Minor 742 Versions", draft-ietf-nfsv4-versioning-09 (work in 743 progress), December 2016. 745 [NSM] The Open Group, "Protocols for Interworking: XNFS, Version 746 3W", February 1998. 748 [RFC1094] Nowicki, B., "NFS: Network File System Protocol 749 specification", RFC 1094, DOI 10.17487/RFC1094, March 750 1989, . 752 [RFC1813] Callaghan, B., Pawlowski, B., and P. Staubach, "NFS 753 Version 3 Protocol Specification", RFC 1813, 754 DOI 10.17487/RFC1813, June 1995, 755 . 757 [RFC5040] Recio, R., Metzler, B., Culley, P., Hilland, J., and D. 758 Garcia, "A Remote Direct Memory Access Protocol 759 Specification", RFC 5040, DOI 10.17487/RFC5040, October 760 2007, . 762 [RFC5041] Shah, H., Pinkerton, J., Recio, R., and P. Culley, "Direct 763 Data Placement over Reliable Transports", RFC 5041, 764 DOI 10.17487/RFC5041, October 2007, 765 . 767 [RFC5666] Talpey, T. and B. Callaghan, "Remote Direct Memory Access 768 Transport for Remote Procedure Call", RFC 5666, 769 DOI 10.17487/RFC5666, January 2010, 770 . 772 [RFC5667] Talpey, T. and B. Callaghan, "Network File System (NFS) 773 Direct Data Placement", RFC 5667, DOI 10.17487/RFC5667, 774 January 2010, . 776 Appendix A. Changes Since RFC 5667 778 Corrections and updates made necessary by new language in 779 [I-D.ietf-nfsv4-rfc5666bis] have been introduced. For example, 780 references to deprecated features of RPC-over-RDMA Version One, such 781 as RDMA_MSGP, and the use of the Read list for handling RPC replies, 782 have been removed. The term "mapping" has been replaced with the 783 term "binding" or "Upper Layer Binding" throughout the document. 784 Some material that duplicates what is in [I-D.ietf-nfsv4-rfc5666bis] 785 has been deleted. 787 Material required by [I-D.ietf-nfsv4-rfc5666bis] for Upper Layer 788 Bindings that was not present in [RFC5667] has been added, including 789 discussion of how each NFS version properly estimates the maximum 790 size of RPC replies. 792 Technical corrections have been made. For example, the mention of 793 12KB and 36KB inline thresholds have been removed. The reference to 794 a non-existant NFS version 4 SYMLINK operation has been replaced. 796 The discussion of NFS version 4 COMPOUND handling has been completed. 797 Some changes were made to the algorithm for matching DDP-eligible 798 results to Write chunks. 800 Requirements to ignore extra Read or Write chunks have been removed 801 from the NFS version 2 and 3 Upper Layer Binding, as they conflict 802 with [I-D.ietf-nfsv4-rfc5666bis]. 804 A complete discussion of reply size estimation has been introduced 805 for all protocols covered by the Upper Layer Bindings in this 806 document. 808 A section discussing NFS version 4 retransmission and connection loss 809 has been added. 811 The following additional improvements have been made, relative to 812 [RFC5667]: 814 o An explicit discussion of NFS version 4.0 and NFS version 4.1 815 backchannel operation has replaced the previous treatment of 816 callback operations. 818 o A binding for NFS version 4.2 has been added that includes 819 discussion of new data-bearing operations like READ_PLUS. 821 o A section suggesting a mechanism for periodically assessing 822 connection health has been introduced. 824 o Language inconsistent with or contradictory to 825 [I-D.ietf-nfsv4-rfc5666bis] has been removed from the present 826 document. 828 o Ambiguous or erroneous uses of RFC2119 terms have been corrected. 830 o References to obsolete RFCs have been updated. 832 o An IANA Considerations Section has been added, which specifies the 833 port assignments for NFS/RDMA. This replaces the example 834 assignment that appeared in [RFC5666]. 836 o Code excerpts have been removed, and figures have been modernized. 838 Appendix B. Acknowledgments 840 The author gratefully acknowledges the work of Brent Callaghan and 841 Tom Talpey on the original NFS Direct Data Placement specification 842 [RFC5667]. The author also wishes to thank Bill Baker and Greg 843 Marsden for their support of this work. 845 Dave Noveck provided excellent review, constructive suggestions, and 846 consistent navigational guidance throughout the process of drafting 847 this document. Dave also contributed the text of Section 5.6 and 848 Section 6, and insisted on precise discussion of reply size 849 estimation. 851 Thanks to Karen Deitke for her sharp observations about idempotency, 852 and the clarity of the discussion of NFS COMPOUNDs and NFS sessions. 854 Special thanks go to Transport Area Director Spencer Dawkins, nfsv4 855 Working Group Chair Spencer Shepler, and nfsv4 Working Group 856 Secretary Thomas Haynes for their support. 858 Author's Address 860 Charles Lever (editor) 861 Oracle Corporation 862 1015 Granger Avenue 863 Ann Arbor, MI 48104 864 USA 866 Phone: +1 248 816 6463 867 Email: chuck.lever@oracle.com