| Network File System Version 4 | C. Lever, Ed. |
| Internet-Draft | Oracle |
| Updates: 7530 (if approved) | D. Noveck |
| Intended status: Standards Track | NetApp |
| Expires: August 5, 2019 | February 1, 2019 |
NFS version 4.0 Trunking Update
draft-ietf-nfsv4-mv0-trunking-update-05
The file system location-related attribute in NFS version 4.0, fs_locations, informs clients about alternate locations of file systems. An NFS version 4.0 client can use this information to handle migration and replication of server filesystems. This document describes how an NFS version 4.0 client can additionally use this information to discover an NFS version 4.0 server's trunking capabilities. This document updates RFC 7530.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on August 5, 2019.
Copyright (c) 2019 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
The NFS version 4.0 specification [RFC7530] defines a migration feature that enables the transfer of a file system from one server to another without disruption of client activity. There were a number of issues with the original definition of this feature, now resolved with the publication of [RFC7931].
After a migration event, a client must determine whether state recovery is necessary. To do this, it needs to determine whether the source and destination server addresses represent the same server instance, if the client has already established a lease on the destination server for other file systems, and if the destination server instance has lock state for the migrated file system.
As part of addressing this need, [RFC7931] introduces trunking into NFS version 4.0 along with a trunking detection mechanism. A trunking detection mechanism enables a client to determine whether two distinct network addresses are connected to the same NFS version 4.0 server instance. Without this knowledge, a client unaware of a trunking relationship between paths it is using simultaneously is likely to become confused in ways described in [RFC7530].
NFSv4.1 was defined with an integral means of trunking detection, described in [RFC5661]. NFSv4.0 initially did not have one, with it being added by [RFC7931]. Nevertheless, the use of the concept of server-trunkability is the same in both protocol versions.
File system migration, replication, and referrals are distinct protocol features. However, it is not appropriate to treat each of these features in isolation. For example, client migration recovery processing needs to deal with the possibility of multiple server addresses in a returned fs_locations attribute. In addition, the contents of the fs_locations attribute, which provides both trunking-related and replication information, may change over repeated retrievals, requiring an integrated description of how clients are to deal with such changes. The issues discussed in the current document relate to the interpretation of the fs_locations attribute and to the proper client and server handling of changes in fs_locations attribute values.
Therefore the goals of the current document are:
The current document pursues these goals by presenting a set of updates to [RFC7530] as summarized in Sections 5 and 6 below.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.
Most of the terms related to handling the fs_locations attribute are appropriately defined in Section 5.1 below. However, there are a few terms used outside that context that are explained in this section.
Trunking refers to a situation in which a client uses multiple network addresses communicate with the same server. Trunking was first introduced to NFSv4.0 by [RFC7931]. Regarding network addresses and the handling of trunking we use the following terminology:
Particularly important is the distinction between trunking detection and trunking discovery. The definitions we present are applicable to all minor versions of NFSv4, but we put particular emphasis on how these terms apply to NFS version 4.0.
Discussion of the term "replica" is complicated for a number of reasons. Even though the term is used in explaining the issues in [RFC7530] that need to be addressed in the current document, a full explanation of this term requires explanation of related terms connected to the fs_locations attribute, which is provided in Section 5.1 of the current document.
The term is also used in previous documents about NFSv4.0 (i.e., [RFC7530] and [RFC7931]) with a meaning different from that in the current document. In these documents each replica is identified by a single network access path. However, in the current document a set of network access paths which have server-trunkable network addresses and the same root-relative file system pathname are considered to be a single replica with multiple network access paths. Although [RFC7931] enables an NFSv4.0 client to determine whether two network addresses were server-trunkable, it never described these as connected to a single replica, leaving in effect the approach established in [RFC7530].
The sections of the current document are divided into four types based on how they relate to the eventual updating of the NFS verion 4.0 specification. Once this update is published, NFS version 4.0 will be specified by multiple documents that need to be read together, until such time as a consolidated replacement specification is produced.
The section types are as follows. See Appendix A for a classification of each section of the current document.
Most of the updates to [RFC7530] to provide support for trunking using the fs_locations attribute apply to Section 8 of that document, entitled "Multi-Server Namespace".
The fs_locations attribute allows specification of file system locations where the data corresponding to a given file system may be accessed. This attribute represents such file system instances as a server address target (as either a DNS host name representing one or more network addresses, or a single literal network address) together with the path of that file system within the associated single-server namespace. Individual fs_locations entries can express trunkable addresses, locations of file system replicas on other servers, migration targets, or pure referrals.
We introduce the following terminology:
Regarding terminology relating to GETATTR attributes used in trunking discovery and other multi-server namespace features:
The subsections below provide replacement sections for existing sections within Section 8.4 of [RFC7530] or new sub-sections to be added to that section.
Together with the possibility of absent file systems, the file system location-bearing attribute fs_locations provides a number of important facilities that enable reliable, manageable, and scalable data access.
When a file system is present on the queried server, this attribute can provide a set of locations that clients may use to access the file system. In the event that server failure, communications problems, or other difficulties make continued access to the file system impossible or otherwise impractical, the returned information provides alternate locations that enable continued access to the file system. Provision of such alternative file system locations is referred to as "replication".
When alternative file system locations are provided, they may represent distinct physical copies of the same file system data or separate NFS server instances that provide access to the same physical file system. Another possible use of the provision of multiple file system location entries is trunking, wherein the file system location entries do not in fact represent different servers but rather are distinct network paths to the same server.
A client may use file system location elements simultaneously to provide higher-performance access to the target file system. This can be done using trunking, although the use of multiple replicas simultaneously is possible. To enable simultaneous access, the client utilizes trunking detection and/or discovery, further described in Section 5.2.2 of the current document, to determine a set of network paths that are server-trunkable with the one currently being used to access the file system. Once this determination is made, requests may be routed across multiple paths, using the existing state management mechanism.
Multiple replicas may also be used simultaneously, typically used when accessing read-only datasets. In this case, each replica requires its own state management. The client performs multiple file opens to read the same file content from multiple replicas.
When a file system is present and subsequently becomes absent, clients can be given the opportunity to have continued access to their data at an alternative file system location. Transfer of the file system contents to the new file system location is referred to as "migration". The client's responsibilities in dealing with this transition depend on the specific nature of the new access path as well as how and whether data was in fact migrated. See Sections 5.2.5 and 5.2.6 of the current document for details.
The fs_locations attribute can designate one or more remote file system locations in place of an absent file system. This is known as a "referral". A particularly important case is that of a "pure referral", in which the absent file system has never been present on the NFS server. Such a referral is a means by which a file system located on one server can redirect clients to file systems located on other servers, thus enabling the creation of a multi-server namespace.
Because client support for the fs_locations attribute is OPTIONAL, a server may (but is not required to) take action to hide migration and referral events from such clients by acting as a proxy, for example.
Trunking is a situation in which multiple distinct network addresses are associated with the same NFS server instance. As a matter of convenience, we say that two network addresses connected to the same NFS server instance are server-trunkable. Section 5.4 of [RFC7931] explains why NFSv4 clients need to be aware of NFS server identity to manage lease and lock state effectively when multiple connections to the same server exist.
Trunking detection refers to a way for an NFSv4 client to confirm that two independently acquired network addresses are connected to the same NFSv4 server. Section 5.8 of [RFC7931] describes an OPTIONAL means by which it can be determined if two network addresses correspond to the same NFSv4.0 server instance. Without trunking detection, an NFSv4.0 client has no other way to confirm that two network addresses are server-trunkable.
In the particular context of NFS version 4.0, trunking detection requires that the client support the Uniform Client ID String approach (UCS), described in Section 5.6 of [RFC7931]. Any NFSv4.0 client that supports migration or trunking detection needs to present a Uniform Client ID String to all NFSv4.0 servers. If it does not do so, it will be unable to perform trunking detection.
Trunking discovery is the process by which an NFSv4 client using a host name or one of an NFSv4 server's network addresses can obtain other candidate network addresses that are trunkable with it; i.e., a set of addresses that might be connected to the same NFSv4 server instance. An NFSv4.0 client can discover server-trunkable network addresses in a number of ways:
When there is a means of trunking detection available, an NFSv4.0 client can confirm that a set of network addresses correspond to the same NFSv4.0 server instance and thus any of them can be used to access that server.
NFS version 4.0 may be implemented using a number of different types of connections:
Because of the need to support multiple connection types, clients face the issue of determining the proper connection type to use when establishing a connection to a server network address. The fs_locations attribute provides no information to support connection type selection. As a result, clients supporting multiple connection types need to attempt to establish a connection on various connection types allowing it to determine, via a trial-and-error approach, which connection types are supported.
If a client strongly prefers one connection type, it can perform these attempts serially in order of declining preference. Once there is a successful attempt, the established connection can be used. Note that with this approach, network partitions can result in a sequence of long waits for a successful connection.
To avoid waiting when there is at least one viable network path available, simultaneous attempts to establish multiple connection types are possible. Once a viable connection is established, the client discards less-preferred connections.
On first access to a file system, the client should obtain the value of the set of alternative file system locations by interrogating the fs_locations attribute. Trunking discovery and/or detection can then be applied to the file system location entries to separate the candidate server-trunkable addresses from the replica addresses that provide alternative locations of the file system. Server-trunkable addresses may be used simultaneously to provide higher performance through the exploitation of multiple paths between client and target file system.
In the event that server failures, communications problems, or other difficulties make continued access to the current file system impossible or otherwise impractical, the client can use the alternative file system locations as a way to maintain continued access to the file system. See Section 5.2.6 of the current document for more detail.
When a file system is present and becomes absent, clients can be given the opportunity to have continued access to their data at an alternative file system location specified by the fs_locations attribute. Typically, a client will be accessing the file system in question, get an NFS4ERR_MOVED error, and then use the fs_locations attribute to determine the new location of the data. See Section 5.2.6 of the current document for more detail.
Such migration can help provide load balancing or general resource reallocation. The protocol does not specify how the file system will be moved between servers. It is anticipated that a number of different server-to-server transfer mechanisms might be used, with the choice left to the server implementer. The NFSv4 protocol specifies the method used to communicate the migration event between client and server.
When the client receives indication of a migration event via an NFS4ERR_MOVED error, data propagation to the destination server must have already occurred. Once the client proceeds to access the alternate file system location, it must see the same data. Where file systems are writable, a change made on the original file system must be visible on all migration targets. Where a file system is not writable but represents a read-only copy (possibly periodically updated) of a writable file system, similar requirements apply to the propagation of updates. Any change visible in the original file system must already be effected on all migration targets, to avoid any possibility that a client, in effecting a transition to the migration target, will see any reversion in file system state.
When the set of network addresses on a server change in a way that would affect a file system location attribute, there are several possible outcomes for clients currently accessing that file system. NFS4ERR_MOVED is returned only when the server cannot satisfy a request from the client, whether because the file system has been migrated to a different server, is only accessible at a different trunked address on the same server, or some other reason. In the cases 1 and 2 below, NFS4ERR_MOVED is not returned.
When migration does occur, multiple addresses may be in use on the server previous to migration and multiple addresses may be available for use on the destination server.
With regard to the server in use, it may be that return of NFS4ERR_MOVED indicates that a particular network address is no longer to be used, without implying that migration of the file system to a different server is needed. Clients should not conclude that migration has occurred until confirming that all network addresses known to be associated with that server are not usable.
It should be noted that the need to defer this determination is not absolute. If a client is not aware of all network addresses for any reason, it may conclude that migration has occurred when it has not and treat a switch to a different server address as if it were a migration event. This is harmless since the use of the same server via a new address will appear as a successful instance of Transparent State Migration.
Although significant harm cannot arise from this misapprehension, it can give rise to disconcerting situations. For example, if a lock has been revoked during the address shift, it will appear to the client as if the lock has been lost during migration. When such a lock is lost, it is the responsibility of the destination server to provide for its recovery via the use of an fs-specific grace period.
With regard to the destination server, it is desirable for the client to be aware of all valid network addresses that can be used to access the destination server. However, there is no need for this to be done immediately. Implementations can process the additional file system location elements in parallel with normal use of the first valid file system location entry found to access the destination.
Because a file system location attribute may include entries relating to the current server, the migration destination, and possible replicas to use, scanning for available network addresses that might be trunkable with addresses the client has already seen could potentially be a long process. To keep this process as short as possible, servers that provide information about trunkable network paths are REQUIRED to place file system location entries that represent addresses usable with the current server or a migration target before those associated with replicas.
This ordering allows a client to cease scanning for trunkable file system location entries once it encounters a file system location element whose fs_name differs from the current fs_name, or whose address is not server-trunkable with the one it is currently using. Although the possibility exists that a client might prematurely cease scanning for trunkable addresses when receiving a location attribute from an older server that does not follow the ordering constraint above, the harm is expected to be limited since such servers would not be expected to present information about trunkable server access paths.
As mentioned above, a single file system location entry may have a server address target in the form of a DNS host name that resolves to multiple network addresses; it is also possible for multiple file system location entries to have their own server address targets that reference the same server.
When server-trunkable addresses for a server exist, the client may assume that for each file system in the namespace of a given server network address, there exist file systems at corresponding namespace locations for each of the other server-trunkable network addresses. It may do this even in the absence of explicit listing in fs_locations. Such corresponding file system locations can be used as alternative locations, just as those explicitly specified via the fs_locations attribute.
If a single file system location entry designates multiple server IP addresses, the client should choose a single one to use. When two server addresses are designated by a single file system location entry and they correspond to different servers, this normally indicates some sort of misconfiguration. The client should avoid using such file system location entries when alternatives are available. When they are not, the client should pick one of the IP addresses and use it, without using others that are not directed to the same server.
Since the existing description of NFS4ERR_MOVED in Section 13.1.2.4 of [RFC7530] does not take proper account of trunking, it needs to be modified by replacing the first two sentences of the description with the following material:
The Security Considerations section of [RFC7530] needs the additions below to properly address some aspects of trunking discovery, referral, migration, and replication.
Privacy considerations relating to uniform client strings (UCS) vs. non-uniform client strings (non-UCS), discussed in Section 5.6 of
[RFC7931], are also applicable to their usage for trunking detection in NFS version 4.0.
This document does not require actions by IANA.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC5531] | Thurlow, R., "RPC: Remote Procedure Call Protocol Specification Version 2", RFC 5531, DOI 10.17487/RFC5531, May 2009. |
| [RFC7530] | Haynes, T. and D. Noveck, "Network File System (NFS) Version 4 Protocol", RFC 7530, DOI 10.17487/RFC7530, March 2015. |
| [RFC7931] | Noveck, D., Shivam, P., Lever, C. and B. Baker, "NFSv4.0 Migration: Specification Update", RFC 7931, DOI 10.17487/RFC7931, July 2016. |
| [RFC8166] | Lever, C., Simpson, W. and T. Talpey, "Remote Direct Memory Access Transport for Remote Procedure Call Version 1", RFC 8166, DOI 10.17487/RFC8166, June 2017. |
| [RFC8174] | Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017. |
| [RFC4033] | Arends, R., Austein, R., Larson, M., Massey, D. and S. Rose, "DNS Security Introduction and Requirements", RFC 4033, DOI 10.17487/RFC4033, March 2005. |
| [RFC5661] | Shepler, S., Eisler, M. and D. Noveck, "Network File System (NFS) Version 4 Minor Version 1 Protocol", RFC 5661, DOI 10.17487/RFC5661, January 2010. |
| [RFC7861] | Adamson, A. and N. Williams, "Remote Procedure Call (RPC) Security Version 3", RFC 7861, DOI 10.17487/RFC7861, November 2016. |
| [RFC8446] | Rescorla, E., "The Transport Layer Security (TLS) Protocol Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018. |
All sections of the current document are considered explanatory with the following exceptions.
The authors wish to thank Andy Adamson, who wrote the original version of this document. All the innovation in this document is the result of Andy's work, while mistakes are best ascribed to the current authors.
The editor wishes to thank Greg Marsden for his support of this work, and Robert Thurlow for his review and suggestions.
Special thanks go to Transport Area Director Spencer Dawkins, NFSV4 Working Group Chairs Spencer Shepler and Brian Pawlowski, and NFSV4 Working Group Secretary Thomas Haynes for their ongoing support. We are also grateful for the thorough review of this document by Benjamin Kaduk and Ben Campbell.