1NFS(5) File Formats Manual NFS(5)
2
6 nfs - fstab format and options for the nfs file systems
7
9 /etc/fstab
10
12 NFS is an Internet Standard protocol created by Sun Microsystems in
13 1984. NFS was developed to allow file sharing between systems residing
14 on a local area network. Depending on kernel configuration, the Linux
15 NFS client may support NFS versions 2, 3, 4.0, 4.1, or 4.2.
16 The mount(8) command attaches a file system to the system's name space
18 hierarchy at a given mount point. The /etc/fstab file describes how
19 mount(8) should assemble a system's file name hierarchy from various
20 independent file systems (including file systems exported by NFS
21 servers). Each line in the /etc/fstab file describes a single file
22 system, its mount point, and a set of default mount options for that
23 mount point.
24 For NFS file system mounts, a line in the /etc/fstab file specifies the
26 server name, the path name of the exported server directory to mount,
27 the local directory that is the mount point, the type of file system
28 that is being mounted, and a list of mount options that control the way
29 the filesystem is mounted and how the NFS client behaves when accessing
30 files on this mount point. The fifth and sixth fields on each line are
31 not used by NFS, thus conventionally each contain the digit zero. For
32 example:
33 server:path /mountpoint fstype option,option,... 0 0
35 The server's hostname and export pathname are separated by a colon,
37 while the mount options are separated by commas. The remaining fields
38 are separated by blanks or tabs.
39 The server's hostname can be an unqualified hostname, a fully qualified
41 domain name, a dotted quad IPv4 address, or an IPv6 address enclosed in
42 square brackets. Link-local and site-local IPv6 addresses must be
43 accompanied by an interface identifier. See ipv6(7) for details on
44 specifying raw IPv6 addresses.
45 The fstype field contains "nfs". Use of the "nfs4" fstype in
47 /etc/fstab is deprecated.
48
50 Refer to mount(8) for a description of generic mount options available
51 for all file systems. If you do not need to specify any mount options,
52 use the generic option defaults in /etc/fstab.
53 Options supported by all versions
55 These options are valid to use with any NFS version.
56 nfsvers=n The NFS protocol version number used to contact the
58 server's NFS service. If the server does not support
59 the requested version, the mount request fails. If this
60 option is not specified, the client tries version 4.2
61 first, then negotiates down until it finds a version
62 supported by the server.
63 vers=n This option is an alternative to the nfsvers option. It
65 is included for compatibility with other operating sys‐
66 tems
67 soft / hard Determines the recovery behavior of the NFS client after
69 an NFS request times out. If neither option is speci‐
70 fied (or if the hard option is specified), NFS requests
71 are retried indefinitely. If the soft option is speci‐
72 fied, then the NFS client fails an NFS request after
73 retrans retransmissions have been sent, causing the NFS
74 client to return an error to the calling application.
75 NB: A so-called "soft" timeout can cause silent data
77 corruption in certain cases. As such, use the soft
78 option only when client responsiveness is more important
79 than data integrity. Using NFS over TCP or increasing
80 the value of the retrans option may mitigate some of the
81 risks of using the soft option.
82 intr / nointr This option is provided for backward compatibility. It
84 is ignored after kernel 2.6.25.
85 timeo=n The time in deciseconds (tenths of a second) the NFS
87 client waits for a response before it retries an NFS
88 request.
89 For NFS over TCP the default timeo value is 600 (60 sec‐
91 onds). The NFS client performs linear backoff: After
92 each retransmission the timeout is increased by timeo up
93 to the maximum of 600 seconds.
94 However, for NFS over UDP, the client uses an adaptive
96 algorithm to estimate an appropriate timeout value for
97 frequently used request types (such as READ and WRITE
98 requests), but uses the timeo setting for infrequently
99 used request types (such as FSINFO requests). If the
100 timeo option is not specified, infrequently used request
101 types are retried after 1.1 seconds. After each
102 retransmission, the NFS client doubles the timeout for
103 that request, up to a maximum timeout length of 60 sec‐
104 onds.
105 retrans=n The number of times the NFS client retries a request
107 before it attempts further recovery action. If the
108 retrans option is not specified, the NFS client tries
109 each UDP request three times and each TCP request twice.
110 The NFS client generates a "server not responding" mes‐
112 sage after retrans retries, then attempts further recov‐
113 ery (depending on whether the hard mount option is in
114 effect).
115 rsize=n The maximum number of bytes in each network READ request
117 that the NFS client can receive when reading data from a
118 file on an NFS server. The actual data payload size of
119 each NFS READ request is equal to or smaller than the
120 rsize setting. The largest read payload supported by the
121 Linux NFS client is 1,048,576 bytes (one megabyte).
122 The rsize value is a positive integral multiple of 1024.
124 Specified rsize values lower than 1024 are replaced with
125 4096; values larger than 1048576 are replaced with
126 1048576. If a specified value is within the supported
127 range but not a multiple of 1024, it is rounded down to
128 the nearest multiple of 1024.
129 If an rsize value is not specified, or if the specified
131 rsize value is larger than the maximum that either
132 client or server can support, the client and server
133 negotiate the largest rsize value that they can both
134 support.
135 The rsize mount option as specified on the mount(8) com‐
137 mand line appears in the /etc/mtab file. However, the
138 effective rsize value negotiated by the client and
139 server is reported in the /proc/mounts file.
140 wsize=n The maximum number of bytes per network WRITE request
142 that the NFS client can send when writing data to a file
143 on an NFS server. The actual data payload size of each
144 NFS WRITE request is equal to or smaller than the wsize
145 setting. The largest write payload supported by the
146 Linux NFS client is 1,048,576 bytes (one megabyte).
147 Similar to rsize , the wsize value is a positive inte‐
149 gral multiple of 1024. Specified wsize values lower
150 than 1024 are replaced with 4096; values larger than
151 1048576 are replaced with 1048576. If a specified value
152 is within the supported range but not a multiple of
153 1024, it is rounded down to the nearest multiple of
154 1024.
155 If a wsize value is not specified, or if the specified
157 wsize value is larger than the maximum that either
158 client or server can support, the client and server
159 negotiate the largest wsize value that they can both
160 support.
161 The wsize mount option as specified on the mount(8) com‐
163 mand line appears in the /etc/mtab file. However, the
164 effective wsize value negotiated by the client and
165 server is reported in the /proc/mounts file.
166 ac / noac Selects whether the client may cache file attributes. If
168 neither option is specified (or if ac is specified), the
169 client caches file attributes.
170 To improve performance, NFS clients cache file
172 attributes. Every few seconds, an NFS client checks the
173 server's version of each file's attributes for updates.
174 Changes that occur on the server in those small inter‐
175 vals remain undetected until the client checks the
176 server again. The noac option prevents clients from
177 caching file attributes so that applications can more
178 quickly detect file changes on the server.
179 In addition to preventing the client from caching file
181 attributes, the noac option forces application writes to
182 become synchronous so that local changes to a file
183 become visible on the server immediately. That way,
184 other clients can quickly detect recent writes when they
185 check the file's attributes.
186 Using the noac option provides greater cache coherence
188 among NFS clients accessing the same files, but it
189 extracts a significant performance penalty. As such,
190 judicious use of file locking is encouraged instead.
191 The DATA AND METADATA COHERENCE section contains a
192 detailed discussion of these trade-offs.
193 acregmin=n The minimum time (in seconds) that the NFS client caches
195 attributes of a regular file before it requests fresh
196 attribute information from a server. If this option is
197 not specified, the NFS client uses a 3-second minimum.
198 See the DATA AND METADATA COHERENCE section for a full
199 discussion of attribute caching.
200 acregmax=n The maximum time (in seconds) that the NFS client caches
202 attributes of a regular file before it requests fresh
203 attribute information from a server. If this option is
204 not specified, the NFS client uses a 60-second maximum.
205 See the DATA AND METADATA COHERENCE section for a full
206 discussion of attribute caching.
207 acdirmin=n The minimum time (in seconds) that the NFS client caches
209 attributes of a directory before it requests fresh
210 attribute information from a server. If this option is
211 not specified, the NFS client uses a 30-second minimum.
212 See the DATA AND METADATA COHERENCE section for a full
213 discussion of attribute caching.
214 acdirmax=n The maximum time (in seconds) that the NFS client caches
216 attributes of a directory before it requests fresh
217 attribute information from a server. If this option is
218 not specified, the NFS client uses a 60-second maximum.
219 See the DATA AND METADATA COHERENCE section for a full
220 discussion of attribute caching.
221 actimeo=n Using actimeo sets all of acregmin, acregmax, acdirmin,
223 and acdirmax to the same value. If this option is not
224 specified, the NFS client uses the defaults for each of
225 these options listed above.
226 bg / fg Determines how the mount(8) command behaves if an
228 attempt to mount an export fails. The fg option causes
229 mount(8) to exit with an error status if any part of the
230 mount request times out or fails outright. This is
231 called a "foreground" mount, and is the default behavior
232 if neither the fg nor bg mount option is specified.
233 If the bg option is specified, a timeout or failure
235 causes the mount(8) command to fork a child which con‐
236 tinues to attempt to mount the export. The parent imme‐
237 diately returns with a zero exit code. This is known as
238 a "background" mount.
239 If the local mount point directory is missing, the
241 mount(8) command acts as if the mount request timed out.
242 This permits nested NFS mounts specified in /etc/fstab
243 to proceed in any order during system initialization,
244 even if some NFS servers are not yet available. Alter‐
245 natively these issues can be addressed using an auto‐
246 mounter (refer to automount(8) for details).
247 rdirplus / nordirplus
249 Selects whether to use NFS v3 or v4 READDIRPLUS
250 requests. If this option is not specified, the NFS
251 client uses READDIRPLUS requests on NFS v3 or v4 mounts
252 to read small directories. Some applications perform
253 better if the client uses only READDIR requests for all
254 directories.
255 retry=n The number of minutes that the mount(8) command retries
257 an NFS mount operation in the foreground or background
258 before giving up. If this option is not specified, the
259 default value for foreground mounts is 2 minutes, and
260 the default value for background mounts is 10000 minutes
261 (80 minutes shy of one week). If a value of zero is
262 specified, the mount(8) command exits immediately after
263 the first failure.
264 Note that this only affects how many retries are made
266 and doesn't affect the delay caused by each retry. For
267 UDP each retry takes the time determined by the timeo
268 and retrans options, which by default will be about 7
269 seconds. For TCP the default is 3 minutes, but system
270 TCP connection timeouts will sometimes limit the timeout
271 of each retransmission to around 2 minutes.
272 sec=flavors A colon-separated list of one or more security flavors
274 to use for accessing files on the mounted export. If the
275 server does not support any of these flavors, the mount
276 operation fails. If sec= is not specified, the client
277 attempts to find a security flavor that both the client
278 and the server supports. Valid flavors are none, sys,
279 krb5, krb5i, and krb5p. Refer to the SECURITY CONSIDER‐
280 ATIONS section for details.
281 sharecache / nosharecache
283 Determines how the client's data cache and attribute
284 cache are shared when mounting the same export more than
285 once concurrently. Using the same cache reduces memory
286 requirements on the client and presents identical file
287 contents to applications when the same remote file is
288 accessed via different mount points.
289 If neither option is specified, or if the sharecache
291 option is specified, then a single cache is used for all
292 mount points that access the same export. If the
293 nosharecache option is specified, then that mount point
294 gets a unique cache. Note that when data and attribute
295 caches are shared, the mount options from the first
296 mount point take effect for subsequent concurrent mounts
297 of the same export.
298 As of kernel 2.6.18, the behavior specified by noshare‐
300 cache is legacy caching behavior. This is considered a
301 data risk since multiple cached copies of the same file
302 on the same client can become out of sync following a
303 local update of one of the copies.
304 resvport / noresvport
306 Specifies whether the NFS client should use a privileged
307 source port when communicating with an NFS server for
308 this mount point. If this option is not specified, or
309 the resvport option is specified, the NFS client uses a
310 privileged source port. If the noresvport option is
311 specified, the NFS client uses a non-privileged source
312 port. This option is supported in kernels 2.6.28 and
313 later.
314 Using non-privileged source ports helps increase the
316 maximum number of NFS mount points allowed on a client,
317 but NFS servers must be configured to allow clients to
318 connect via non-privileged source ports.
319 Refer to the SECURITY CONSIDERATIONS section for impor‐
321 tant details.
322 lookupcache=mode
324 Specifies how the kernel manages its cache of directory
325 entries for a given mount point. mode can be one of
326 all, none, pos, or positive. This option is supported
327 in kernels 2.6.28 and later.
328 The Linux NFS client caches the result of all NFS LOOKUP
330 requests. If the requested directory entry exists on
331 the server, the result is referred to as positive. If
332 the requested directory entry does not exist on the
333 server, the result is referred to as negative.
334 If this option is not specified, or if all is specified,
336 the client assumes both types of directory cache entries
337 are valid until their parent directory's cached
338 attributes expire.
339 If pos or positive is specified, the client assumes pos‐
341 itive entries are valid until their parent directory's
342 cached attributes expire, but always revalidates nega‐
343 tive entires before an application can use them.
344 If none is specified, the client revalidates both types
346 of directory cache entries before an application can use
347 them. This permits quick detection of files that were
348 created or removed by other clients, but can impact
349 application and server performance.
350 The DATA AND METADATA COHERENCE section contains a
352 detailed discussion of these trade-offs.
353 fsc / nofsc Enable/Disables the cache of (read-only) data pages to
355 the local disk using the FS-Cache facility. See
356 cachefilesd(8) and <kernel_soruce>/Documenta‐
357 tion/filesystems/caching for detail on how to configure
358 the FS-Cache facility. Default value is nofsc.
359 Options for NFS versions 2 and 3 only
361 Use these options, along with the options in the above subsection, for
362 NFS versions 2 and 3 only.
363 proto=netid The netid determines the transport that is used to com‐
365 municate with the NFS server. Available options are
366 udp, udp6, tcp, tcp6, and rdma. Those which end in 6
367 use IPv6 addresses and are only available if support for
368 TI-RPC is built in. Others use IPv4 addresses.
369 Each transport protocol uses different default retrans
371 and timeo settings. Refer to the description of these
372 two mount options for details.
373 In addition to controlling how the NFS client transmits
375 requests to the server, this mount option also controls
376 how the mount(8) command communicates with the server's
377 rpcbind and mountd services. Specifying a netid that
378 uses TCP forces all traffic from the mount(8) command
379 and the NFS client to use TCP. Specifying a netid that
380 uses UDP forces all traffic types to use UDP.
381 Before using NFS over UDP, refer to the TRANSPORT METH‐
383 ODS section.
384 If the proto mount option is not specified, the mount(8)
386 command discovers which protocols the server supports
387 and chooses an appropriate transport for each service.
388 Refer to the TRANSPORT METHODS section for more details.
389 udp The udp option is an alternative to specifying
391 proto=udp. It is included for compatibility with other
392 operating systems.
393 Before using NFS over UDP, refer to the TRANSPORT METH‐
395 ODS section.
396 tcp The tcp option is an alternative to specifying
398 proto=tcp. It is included for compatibility with other
399 operating systems.
400 rdma The rdma option is an alternative to specifying
402 proto=rdma.
403 port=n The numeric value of the server's NFS service port. If
405 the server's NFS service is not available on the speci‐
406 fied port, the mount request fails.
407 If this option is not specified, or if the specified
409 port value is 0, then the NFS client uses the NFS ser‐
410 vice port number advertised by the server's rpcbind ser‐
411 vice. The mount request fails if the server's rpcbind
412 service is not available, the server's NFS service is
413 not registered with its rpcbind service, or the server's
414 NFS service is not available on the advertised port.
415 mountport=n The numeric value of the server's mountd port. If the
417 server's mountd service is not available on the speci‐
418 fied port, the mount request fails.
419 If this option is not specified, or if the specified
421 port value is 0, then the mount(8) command uses the
422 mountd service port number advertised by the server's
423 rpcbind service. The mount request fails if the
424 server's rpcbind service is not available, the server's
425 mountd service is not registered with its rpcbind ser‐
426 vice, or the server's mountd service is not available on
427 the advertised port.
428 This option can be used when mounting an NFS server
430 through a firewall that blocks the rpcbind protocol.
431 mountproto=netid
433 The transport the NFS client uses to transmit requests
434 to the NFS server's mountd service when performing this
435 mount request, and when later unmounting this mount
436 point.
437 netid may be one of udp, and tcp which use IPv4 address
439 or, if TI-RPC is built into the mount.nfs command, udp6,
440 and tcp6 which use IPv6 addresses.
441 This option can be used when mounting an NFS server
443 through a firewall that blocks a particular transport.
444 When used in combination with the proto option, differ‐
445 ent transports for mountd requests and NFS requests can
446 be specified. If the server's mountd service is not
447 available via the specified transport, the mount request
448 fails.
449 Refer to the TRANSPORT METHODS section for more on how
451 the mountproto mount option interacts with the proto
452 mount option.
453 mounthost=name The hostname of the host running mountd. If this option
455 is not specified, the mount(8) command assumes that the
456 mountd service runs on the same host as the NFS service.
457 mountvers=n The RPC version number used to contact the server's
459 mountd. If this option is not specified, the client
460 uses a version number appropriate to the requested NFS
461 version. This option is useful when multiple NFS ser‐
462 vices are running on the same remote server host.
463 namlen=n The maximum length of a pathname component on this
465 mount. If this option is not specified, the maximum
466 length is negotiated with the server. In most cases,
467 this maximum length is 255 characters.
468 Some early versions of NFS did not support this negotia‐
470 tion. Using this option ensures that pathconf(3)
471 reports the proper maximum component length to applica‐
472 tions in such cases.
473 lock / nolock Selects whether to use the NLM sideband protocol to lock
475 files on the server. If neither option is specified (or
476 if lock is specified), NLM locking is used for this
477 mount point. When using the nolock option, applications
478 can lock files, but such locks provide exclusion only
479 against other applications running on the same client.
480 Remote applications are not affected by these locks.
481 NLM locking must be disabled with the nolock option when
483 using NFS to mount /var because /var contains files used
484 by the NLM implementation on Linux. Using the nolock
485 option is also required when mounting exports on NFS
486 servers that do not support the NLM protocol.
487 cto / nocto Selects whether to use close-to-open cache coherence
489 semantics. If neither option is specified (or if cto is
490 specified), the client uses close-to-open cache coher‐
491 ence semantics. If the nocto option is specified, the
492 client uses a non-standard heuristic to determine when
493 files on the server have changed.
494 Using the nocto option may improve performance for read-
496 only mounts, but should be used only if the data on the
497 server changes only occasionally. The DATA AND METADATA
498 COHERENCE section discusses the behavior of this option
499 in more detail.
500 acl / noacl Selects whether to use the NFSACL sideband protocol on
502 this mount point. The NFSACL sideband protocol is a
503 proprietary protocol implemented in Solaris that manages
504 Access Control Lists. NFSACL was never made a standard
505 part of the NFS protocol specification.
506 If neither acl nor noacl option is specified, the NFS
508 client negotiates with the server to see if the NFSACL
509 protocol is supported, and uses it if the server sup‐
510 ports it. Disabling the NFSACL sideband protocol may be
511 necessary if the negotiation causes problems on the
512 client or server. Refer to the SECURITY CONSIDERATIONS
513 section for more details.
514 local_lock=mechanism
516 Specifies whether to use local locking for any or both
517 of the flock and the POSIX locking mechanisms. mecha‐
518 nism can be one of all, flock, posix, or none. This
519 option is supported in kernels 2.6.37 and later.
520 The Linux NFS client provides a way to make locks local.
522 This means, the applications can lock files, but such
523 locks provide exclusion only against other applications
524 running on the same client. Remote applications are not
525 affected by these locks.
526 If this option is not specified, or if none is speci‐
528 fied, the client assumes that the locks are not local.
529 If all is specified, the client assumes that both flock
531 and POSIX locks are local.
532 If flock is specified, the client assumes that only
534 flock locks are local and uses NLM sideband protocol to
535 lock files when POSIX locks are used.
536 If posix is specified, the client assumes that POSIX
538 locks are local and uses NLM sideband protocol to lock
539 files when flock locks are used.
540 To support legacy flock behavior similar to that of NFS
542 clients < 2.6.12, use 'local_lock=flock'. This option is
543 required when exporting NFS mounts via Samba as Samba
544 maps Windows share mode locks as flock. Since NFS
545 clients > 2.6.12 implement flock by emulating POSIX
546 locks, this will result in conflicting locks.
547 NOTE: When used together, the 'local_lock' mount option
549 will be overridden by 'nolock'/'lock' mount option.
550 Options for NFS version 4 only
552 Use these options, along with the options in the first subsection
553 above, for NFS version 4.0 and newer.
554 proto=netid The netid determines the transport that is used to com‐
556 municate with the NFS server. Supported options are
557 tcp, tcp6, and rdma. tcp6 use IPv6 addresses and is
558 only available if support for TI-RPC is built in. Both
559 others use IPv4 addresses.
560 All NFS version 4 servers are required to support TCP,
562 so if this mount option is not specified, the NFS ver‐
563 sion 4 client uses the TCP protocol. Refer to the
564 TRANSPORT METHODS section for more details.
565 minorversion=n Specifies the protocol minor version number. NFSv4
567 introduces "minor versioning," where NFS protocol
568 enhancements can be introduced without bumping the NFS
569 protocol version number. Before kernel 2.6.38, the
570 minor version is always zero, and this option is not
571 recognized. After this kernel, specifying "minorver‐
572 sion=1" enables a number of advanced features, such as
573 NFSv4 sessions.
574 Recent kernels allow the minor version to be specified
576 using the vers= option. For example, specifying
577 vers=4.1 is the same as specifying vers=4,minorver‐
578 sion=1.
579 port=n The numeric value of the server's NFS service port. If
581 the server's NFS service is not available on the speci‐
582 fied port, the mount request fails.
583 If this mount option is not specified, the NFS client
585 uses the standard NFS port number of 2049 without first
586 checking the server's rpcbind service. This allows an
587 NFS version 4 client to contact an NFS version 4 server
588 through a firewall that may block rpcbind requests.
589 If the specified port value is 0, then the NFS client
591 uses the NFS service port number advertised by the
592 server's rpcbind service. The mount request fails if
593 the server's rpcbind service is not available, the
594 server's NFS service is not registered with its rpcbind
595 service, or the server's NFS service is not available on
596 the advertised port.
597 cto / nocto Selects whether to use close-to-open cache coherence
599 semantics for NFS directories on this mount point. If
600 neither cto nor nocto is specified, the default is to
601 use close-to-open cache coherence semantics for directo‐
602 ries.
603 File data caching behavior is not affected by this
605 option. The DATA AND METADATA COHERENCE section dis‐
606 cusses the behavior of this option in more detail.
607 clientaddr=n.n.n.n
609 clientaddr=n:n:...:n
611 Specifies a single IPv4 address (in dotted-quad form),
612 or a non-link-local IPv6 address, that the NFS client
613 advertises to allow servers to perform NFS version 4.0
614 callback requests against files on this mount point. If
615 the server is unable to establish callback connections
616 to clients, performance may degrade, or accesses to
617 files may temporarily hang. Can specify a value of
618 IPv4_ANY (0.0.0.0) or equivalent IPv6 any address which
619 will signal to the NFS server that this NFS client does
620 not want delegations.
621 If this option is not specified, the mount(8) command
623 attempts to discover an appropriate callback address
624 automatically. The automatic discovery process is not
625 perfect, however. In the presence of multiple client
626 network interfaces, special routing policies, or atypi‐
627 cal network topologies, the exact address to use for
628 callbacks may be nontrivial to determine.
629 NFS protocol versions 4.1 and 4.2 use the client-estab‐
631 lished TCP connection for callback requests, so do not
632 require the server to connect to the client. This
633 option is therefore only affect NFS version 4.0 mounts.
634 migration / nomigration
636 Selects whether the client uses an identification string
637 that is compatible with NFSv4 Transparent State Migra‐
638 tion (TSM). If the mounted server supports NFSv4 migra‐
639 tion with TSM, specify the migration option.
640 Some server features misbehave in the face of a migra‐
642 tion-compatible identification string. The nomigration
643 option retains the use of a traditional client indenti‐
644 fication string which is compatible with legacy NFS
645 servers. This is also the behavior if neither option is
646 specified. A client's open and lock state cannot be
647 migrated transparently when it identifies itself via a
648 traditional identification string.
649 This mount option has no effect with NFSv4 minor ver‐
651 sions newer than zero, which always use TSM-compatible
652 client identification strings.
653
655 The nfs4 file system type is an old syntax for specifying NFSv4 usage.
656 It can still be used with all NFSv4-specific and common options,
657 excepted the nfsvers mount option.
658
660 If the mount command is configured to do so, all of the mount options
661 described in the previous section can also be configured in the
662 /etc/nfsmount.conf file. See nfsmount.conf(5) for details.
663
665 To mount an export using NFS version 2, use the nfs file system type
666 and specify the nfsvers=2 mount option. To mount using NFS version 3,
667 use the nfs file system type and specify the nfsvers=3 mount option.
668 To mount using NFS version 4, use either the nfs file system type, with
669 the nfsvers=4 mount option, or the nfs4 file system type.
670 The following example from an /etc/fstab file causes the mount command
672 to negotiate reasonable defaults for NFS behavior.
673 server:/export /mnt nfs defaults 0 0
675 Here is an example from an /etc/fstab file for an NFS version 2 mount
677 over UDP.
678 server:/export /mnt nfs nfsvers=2,proto=udp 0 0
680 This example shows how to mount using NFS version 4 over TCP with Ker‐
682 beros 5 mutual authentication.
683 server:/export /mnt nfs4 sec=krb5 0 0
685 This example shows how to mount using NFS version 4 over TCP with Ker‐
687 beros 5 privacy or data integrity mode.
688 server:/export /mnt nfs4 sec=krb5p:krb5i 0 0
690 This example can be used to mount /usr over NFS.
692 server:/export /usr nfs ro,nolock,nocto,actimeo=3600 0 0
694 This example shows how to mount an NFS server using a raw IPv6 link-
696 local address.
697 [fe80::215:c5ff:fb3e:e2b1%eth0]:/export /mnt nfs defaults 0 0
699
701 NFS clients send requests to NFS servers via Remote Procedure Calls, or
702 RPCs. The RPC client discovers remote service endpoints automatically,
703 handles per-request authentication, adjusts request parameters for dif‐
704 ferent byte endianness on client and server, and retransmits requests
705 that may have been lost by the network or server. RPC requests and
706 replies flow over a network transport.
707 In most cases, the mount(8) command, NFS client, and NFS server can
709 automatically negotiate proper transport and data transfer size set‐
710 tings for a mount point. In some cases, however, it pays to specify
711 these settings explicitly using mount options.
712 Traditionally, NFS clients used the UDP transport exclusively for
714 transmitting requests to servers. Though its implementation is simple,
715 NFS over UDP has many limitations that prevent smooth operation and
716 good performance in some common deployment environments. Even an
717 insignificant packet loss rate results in the loss of whole NFS
718 requests; as such, retransmit timeouts are usually in the subsecond
719 range to allow clients to recover quickly from dropped requests, but
720 this can result in extraneous network traffic and server load.
721 However, UDP can be quite effective in specialized settings where the
723 networks MTU is large relative to NFSs data transfer size (such as net‐
724 work environments that enable jumbo Ethernet frames). In such environ‐
725 ments, trimming the rsize and wsize settings so that each NFS read or
726 write request fits in just a few network frames (or even in a single
727 frame) is advised. This reduces the probability that the loss of a
728 single MTU-sized network frame results in the loss of an entire large
729 read or write request.
730 TCP is the default transport protocol used for all modern NFS implemen‐
732 tations. It performs well in almost every conceivable network environ‐
733 ment and provides excellent guarantees against data corruption caused
734 by network unreliability. TCP is often a requirement for mounting a
735 server through a network firewall.
736 Under normal circumstances, networks drop packets much more frequently
738 than NFS servers drop requests. As such, an aggressive retransmit
739 timeout setting for NFS over TCP is unnecessary. Typical timeout set‐
740 tings for NFS over TCP are between one and ten minutes. After the
741 client exhausts its retransmits (the value of the retrans mount
742 option), it assumes a network partition has occurred, and attempts to
743 reconnect to the server on a fresh socket. Since TCP itself makes net‐
744 work data transfer reliable, rsize and wsize can safely be allowed to
745 default to the largest values supported by both client and server,
746 independent of the network's MTU size.
747 Using the mountproto mount option
749 This section applies only to NFS version 2 and version 3 mounts since
750 NFS version 4 does not use a separate protocol for mount requests.
751 The Linux NFS client can use a different transport for contacting an
753 NFS server's rpcbind service, its mountd service, its Network Lock Man‐
754 ager (NLM) service, and its NFS service. The exact transports employed
755 by the Linux NFS client for each mount point depends on the settings of
756 the transport mount options, which include proto, mountproto, udp, and
757 tcp.
758 The client sends Network Status Manager (NSM) notifications via UDP no
760 matter what transport options are specified, but listens for server NSM
761 notifications on both UDP and TCP. The NFS Access Control List
762 (NFSACL) protocol shares the same transport as the main NFS service.
763 If no transport options are specified, the Linux NFS client uses UDP to
765 contact the server's mountd service, and TCP to contact its NLM and NFS
766 services by default.
767 If the server does not support these transports for these services, the
769 mount(8) command attempts to discover what the server supports, and
770 then retries the mount request once using the discovered transports.
771 If the server does not advertise any transport supported by the client
772 or is misconfigured, the mount request fails. If the bg option is in
773 effect, the mount command backgrounds itself and continues to attempt
774 the specified mount request.
775 When the proto option, the udp option, or the tcp option is specified
777 but the mountproto option is not, the specified transport is used to
778 contact both the server's mountd service and for the NLM and NFS ser‐
779 vices.
780 If the mountproto option is specified but none of the proto, udp or tcp
782 options are specified, then the specified transport is used for the
783 initial mountd request, but the mount command attempts to discover what
784 the server supports for the NFS protocol, preferring TCP if both trans‐
785 ports are supported.
786 If both the mountproto and proto (or udp or tcp) options are specified,
788 then the transport specified by the mountproto option is used for the
789 initial mountd request, and the transport specified by the proto option
790 (or the udp or tcp options) is used for NFS, no matter what order these
791 options appear. No automatic service discovery is performed if these
792 options are specified.
793 If any of the proto, udp, tcp, or mountproto options are specified more
795 than once on the same mount command line, then the value of the right‐
796 most instance of each of these options takes effect.
797 Using NFS over UDP on high-speed links
799 Using NFS over UDP on high-speed links such as Gigabit can cause silent
800 data corruption.
801 The problem can be triggered at high loads, and is caused by problems
803 in IP fragment reassembly. NFS read and writes typically transmit UDP
804 packets of 4 Kilobytes or more, which have to be broken up into several
805 fragments in order to be sent over the Ethernet link, which limits
806 packets to 1500 bytes by default. This process happens at the IP net‐
807 work layer and is called fragmentation.
808 In order to identify fragments that belong together, IP assigns a 16bit
810 IP ID value to each packet; fragments generated from the same UDP
811 packet will have the same IP ID. The receiving system will collect
812 these fragments and combine them to form the original UDP packet. This
813 process is called reassembly. The default timeout for packet reassembly
814 is 30 seconds; if the network stack does not receive all fragments of a
815 given packet within this interval, it assumes the missing fragment(s)
816 got lost and discards those it already received.
817 The problem this creates over high-speed links is that it is possible
819 to send more than 65536 packets within 30 seconds. In fact, with heavy
820 NFS traffic one can observe that the IP IDs repeat after about 5 sec‐
821 onds.
822 This has serious effects on reassembly: if one fragment gets lost,
824 another fragment from a different packet but with the same IP ID will
825 arrive within the 30 second timeout, and the network stack will combine
826 these fragments to form a new packet. Most of the time, network layers
827 above IP will detect this mismatched reassembly - in the case of UDP,
828 the UDP checksum, which is a 16 bit checksum over the entire packet
829 payload, will usually not match, and UDP will discard the bad packet.
830 However, the UDP checksum is 16 bit only, so there is a chance of 1 in
832 65536 that it will match even if the packet payload is completely ran‐
833 dom (which very often isn't the case). If that is the case, silent data
834 corruption will occur.
835 This potential should be taken seriously, at least on Gigabit Ethernet.
837 Network speeds of 100Mbit/s should be considered less problematic,
838 because with most traffic patterns IP ID wrap around will take much
839 longer than 30 seconds.
840 It is therefore strongly recommended to use NFS over TCP where possi‐
842 ble, since TCP does not perform fragmentation.
843 If you absolutely have to use NFS over UDP over Gigabit Ethernet, some
845 steps can be taken to mitigate the problem and reduce the probability
846 of corruption:
847 Jumbo frames: Many Gigabit network cards are capable of transmitting
849 frames bigger than the 1500 byte limit of traditional
850 Ethernet, typically 9000 bytes. Using jumbo frames of
851 9000 bytes will allow you to run NFS over UDP at a page
852 size of 8K without fragmentation. Of course, this is
853 only feasible if all involved stations support jumbo
854 frames.
855 To enable a machine to send jumbo frames on cards that
857 support it, it is sufficient to configure the interface
858 for a MTU value of 9000.
859 Lower reassembly timeout:
861 By lowering this timeout below the time it takes the IP
862 ID counter to wrap around, incorrect reassembly of frag‐
863 ments can be prevented as well. To do so, simply write
864 the new timeout value (in seconds) to the file
865 /proc/sys/net/ipv4/ipfrag_time.
866 A value of 2 seconds will greatly reduce the probability
868 of IPID clashes on a single Gigabit link, while still
869 allowing for a reasonable timeout when receiving frag‐
870 mented traffic from distant peers.
871
873 Some modern cluster file systems provide perfect cache coherence among
874 their clients. Perfect cache coherence among disparate NFS clients is
875 expensive to achieve, especially on wide area networks. As such, NFS
876 settles for weaker cache coherence that satisfies the requirements of
877 most file sharing types.
878 Close-to-open cache consistency
880 Typically file sharing is completely sequential. First client A opens
881 a file, writes something to it, then closes it. Then client B opens
882 the same file, and reads the changes.
883 When an application opens a file stored on an NFS version 3 server, the
885 NFS client checks that the file exists on the server and is permitted
886 to the opener by sending a GETATTR or ACCESS request. The NFS client
887 sends these requests regardless of the freshness of the file's cached
888 attributes.
889 When the application closes the file, the NFS client writes back any
891 pending changes to the file so that the next opener can view the
892 changes. This also gives the NFS client an opportunity to report write
893 errors to the application via the return code from close(2).
894 The behavior of checking at open time and flushing at close time is
896 referred to as close-to-open cache consistency, or CTO. It can be dis‐
897 abled for an entire mount point using the nocto mount option.
898 Weak cache consistency
900 There are still opportunities for a client's data cache to contain
901 stale data. The NFS version 3 protocol introduced "weak cache consis‐
902 tency" (also known as WCC) which provides a way of efficiently checking
903 a file's attributes before and after a single request. This allows a
904 client to help identify changes that could have been made by other
905 clients.
906 When a client is using many concurrent operations that update the same
908 file at the same time (for example, during asynchronous write behind),
909 it is still difficult to tell whether it was that client's updates or
910 some other client's updates that altered the file.
911 Attribute caching
913 Use the noac mount option to achieve attribute cache coherence among
914 multiple clients. Almost every file system operation checks file
915 attribute information. The client keeps this information cached for a
916 period of time to reduce network and server load. When noac is in
917 effect, a client's file attribute cache is disabled, so each operation
918 that needs to check a file's attributes is forced to go back to the
919 server. This permits a client to see changes to a file very quickly,
920 at the cost of many extra network operations.
921 Be careful not to confuse the noac option with "no data caching." The
923 noac mount option prevents the client from caching file metadata, but
924 there are still races that may result in data cache incoherence between
925 client and server.
926 The NFS protocol is not designed to support true cluster file system
928 cache coherence without some type of application serialization. If
929 absolute cache coherence among clients is required, applications should
930 use file locking. Alternatively, applications can also open their files
931 with the O_DIRECT flag to disable data caching entirely.
932 File timestamp maintenance
934 NFS servers are responsible for managing file and directory timestamps
935 (atime, ctime, and mtime). When a file is accessed or updated on an
936 NFS server, the file's timestamps are updated just like they would be
937 on a filesystem local to an application.
938 NFS clients cache file attributes, including timestamps. A file's
940 timestamps are updated on NFS clients when its attributes are retrieved
941 from the NFS server. Thus there may be some delay before timestamp
942 updates on an NFS server appear to applications on NFS clients.
943 To comply with the POSIX filesystem standard, the Linux NFS client
945 relies on NFS servers to keep a file's mtime and ctime timestamps prop‐
946 erly up to date. It does this by flushing local data changes to the
947 server before reporting mtime to applications via system calls such as
948 stat(2).
949 The Linux client handles atime updates more loosely, however. NFS
951 clients maintain good performance by caching data, but that means that
952 application reads, which normally update atime, are not reflected to
953 the server where a file's atime is actually maintained.
954 Because of this caching behavior, the Linux NFS client does not support
956 generic atime-related mount options. See mount(8) for details on these
957 options.
958 In particular, the atime/noatime, diratime/nodiratime, relatime/norela‐
960 time, and strictatime/nostrictatime mount options have no effect on NFS
961 mounts.
962 /proc/mounts may report that the relatime mount option is set on NFS
964 mounts, but in fact the atime semantics are always as described here,
965 and are not like relatime semantics.
966 Directory entry caching
968 The Linux NFS client caches the result of all NFS LOOKUP requests. If
969 the requested directory entry exists on the server, the result is
970 referred to as a positive lookup result. If the requested directory
971 entry does not exist on the server (that is, the server returned
972 ENOENT), the result is referred to as negative lookup result.
973 To detect when directory entries have been added or removed on the
975 server, the Linux NFS client watches a directory's mtime. If the
976 client detects a change in a directory's mtime, the client drops all
977 cached LOOKUP results for that directory. Since the directory's mtime
978 is a cached attribute, it may take some time before a client notices it
979 has changed. See the descriptions of the acdirmin, acdirmax, and noac
980 mount options for more information about how long a directory's mtime
981 is cached.
982 Caching directory entries improves the performance of applications that
984 do not share files with applications on other clients. Using cached
985 information about directories can interfere with applications that run
986 concurrently on multiple clients and need to detect the creation or
987 removal of files quickly, however. The lookupcache mount option allows
988 some tuning of directory entry caching behavior.
989 Before kernel release 2.6.28, the Linux NFS client tracked only posi‐
991 tive lookup results. This permitted applications to detect new direc‐
992 tory entries created by other clients quickly while still providing
993 some of the performance benefits of caching. If an application depends
994 on the previous lookup caching behavior of the Linux NFS client, you
995 can use lookupcache=positive.
996 If the client ignores its cache and validates every application lookup
998 request with the server, that client can immediately detect when a new
999 directory entry has been either created or removed by another client.
1000 You can specify this behavior using lookupcache=none. The extra NFS
1001 requests needed if the client does not cache directory entries can
1002 exact a performance penalty. Disabling lookup caching should result in
1003 less of a performance penalty than using noac, and has no effect on how
1004 the NFS client caches the attributes of files.
1005 The sync mount option
1007 The NFS client treats the sync mount option differently than some other
1008 file systems (refer to mount(8) for a description of the generic sync
1009 and async mount options). If neither sync nor async is specified (or
1010 if the async option is specified), the NFS client delays sending appli‐
1011 cation writes to the server until any of these events occur:
1012 Memory pressure forces reclamation of system memory resources.
1014 An application flushes file data explicitly with sync(2),
1016 msync(2), or fsync(3).
1017 An application closes a file with close(2).
1019 The file is locked/unlocked via fcntl(2).
1021 In other words, under normal circumstances, data written by an applica‐
1023 tion may not immediately appear on the server that hosts the file.
1024 If the sync option is specified on a mount point, any system call that
1026 writes data to files on that mount point causes that data to be flushed
1027 to the server before the system call returns control to user space.
1028 This provides greater data cache coherence among clients, but at a sig‐
1029 nificant performance cost.
1030 Applications can use the O_SYNC open flag to force application writes
1032 to individual files to go to the server immediately without the use of
1033 the sync mount option.
1034 Using file locks with NFS
1036 The Network Lock Manager protocol is a separate sideband protocol used
1037 to manage file locks in NFS version 2 and version 3. To support lock
1038 recovery after a client or server reboot, a second sideband protocol --
1039 known as the Network Status Manager protocol -- is also required. In
1040 NFS version 4, file locking is supported directly in the main NFS pro‐
1041 tocol, and the NLM and NSM sideband protocols are not used.
1042 In most cases, NLM and NSM services are started automatically, and no
1044 extra configuration is required. Configure all NFS clients with fully-
1045 qualified domain names to ensure that NFS servers can find clients to
1046 notify them of server reboots.
1047 NLM supports advisory file locks only. To lock NFS files, use fcntl(2)
1049 with the F_GETLK and F_SETLK commands. The NFS client converts file
1050 locks obtained via flock(2) to advisory locks.
1051 When mounting servers that do not support the NLM protocol, or when
1053 mounting an NFS server through a firewall that blocks the NLM service
1054 port, specify the nolock mount option. NLM locking must be disabled
1055 with the nolock option when using NFS to mount /var because /var con‐
1056 tains files used by the NLM implementation on Linux.
1057 Specifying the nolock option may also be advised to improve the perfor‐
1059 mance of a proprietary application which runs on a single client and
1060 uses file locks extensively.
1061 NFS version 4 caching features
1063 The data and metadata caching behavior of NFS version 4 clients is sim‐
1064 ilar to that of earlier versions. However, NFS version 4 adds two fea‐
1065 tures that improve cache behavior: change attributes and file delega‐
1066 tion.
1067 The change attribute is a new part of NFS file and directory metadata
1069 which tracks data changes. It replaces the use of a file's modifica‐
1070 tion and change time stamps as a way for clients to validate the con‐
1071 tent of their caches. Change attributes are independent of the time
1072 stamp resolution on either the server or client, however.
1073 A file delegation is a contract between an NFS version 4 client and
1075 server that allows the client to treat a file temporarily as if no
1076 other client is accessing it. The server promises to notify the client
1077 (via a callback request) if another client attempts to access that
1078 file. Once a file has been delegated to a client, the client can cache
1079 that file's data and metadata aggressively without contacting the
1080 server.
1081 File delegations come in two flavors: read and write. A read delega‐
1083 tion means that the server notifies the client about any other clients
1084 that want to write to the file. A write delegation means that the
1085 client gets notified about either read or write accessors.
1086 Servers grant file delegations when a file is opened, and can recall
1088 delegations at any time when another client wants access to the file
1089 that conflicts with any delegations already granted. Delegations on
1090 directories are not supported.
1091 In order to support delegation callback, the server checks the network
1093 return path to the client during the client's initial contact with the
1094 server. If contact with the client cannot be established, the server
1095 simply does not grant any delegations to that client.
1096
1098 NFS servers control access to file data, but they depend on their RPC
1099 implementation to provide authentication of NFS requests. Traditional
1100 NFS access control mimics the standard mode bit access control provided
1101 in local file systems. Traditional RPC authentication uses a number to
1102 represent each user (usually the user's own uid), a number to represent
1103 the user's group (the user's gid), and a set of up to 16 auxiliary
1104 group numbers to represent other groups of which the user may be a mem‐
1105 ber.
1106 Typically, file data and user ID values appear unencrypted (i.e. "in
1108 the clear") on the network. Moreover, NFS versions 2 and 3 use sepa‐
1109 rate sideband protocols for mounting, locking and unlocking files, and
1110 reporting system status of clients and servers. These auxiliary proto‐
1111 cols use no authentication.
1112 In addition to combining these sideband protocols with the main NFS
1114 protocol, NFS version 4 introduces more advanced forms of access con‐
1115 trol, authentication, and in-transit data protection. The NFS version
1116 4 specification mandates support for strong authentication and security
1117 flavors that provide per-RPC integrity checking and encryption.
1118 Because NFS version 4 combines the function of the sideband protocols
1119 into the main NFS protocol, the new security features apply to all NFS
1120 version 4 operations including mounting, file locking, and so on.
1121 RPCGSS authentication can also be used with NFS versions 2 and 3, but
1122 it does not protect their sideband protocols.
1123 The sec mount option specifies the security flavor used for operations
1125 on behalf of users on that NFS mount point. Specifying sec=krb5 pro‐
1126 vides cryptographic proof of a user's identity in each RPC request.
1127 This provides strong verification of the identity of users accessing
1128 data on the server. Note that additional configuration besides adding
1129 this mount option is required in order to enable Kerberos security.
1130 Refer to the rpc.gssd(8) man page for details.
1131 Two additional flavors of Kerberos security are supported: krb5i and
1133 krb5p. The krb5i security flavor provides a cryptographically strong
1134 guarantee that the data in each RPC request has not been tampered with.
1135 The krb5p security flavor encrypts every RPC request to prevent data
1136 exposure during network transit; however, expect some performance
1137 impact when using integrity checking or encryption. Similar support
1138 for other forms of cryptographic security is also available.
1139 NFS version 4 filesystem crossing
1141 The NFS version 4 protocol allows a client to renegotiate the security
1142 flavor when the client crosses into a new filesystem on the server.
1143 The newly negotiated flavor effects only accesses of the new filesys‐
1144 tem.
1145 Such negotiation typically occurs when a client crosses from a server's
1147 pseudo-fs into one of the server's exported physical filesystems, which
1148 often have more restrictive security settings than the pseudo-fs.
1149 NFS version 4 Leases
1151 In NFS version 4, a lease is a period of time during which a server
1152 irrevocably grants a file lock to a client. If the lease expires, the
1153 server is allowed to revoke that lock. Clients periodically renew
1154 their leases to prevent lock revocation.
1155 After an NFS version 4 server reboots, each client tells the server
1157 about all file open and lock state under its lease before operation can
1158 continue. If the client reboots, the server frees all open and lock
1159 state associated with that client's lease.
1160 As part of establishing a lease, therefore, a client must identify
1162 itself to a server. A fixed string is used to distinguish that client
1163 from others, and a changeable verifier is used to indicate when the
1164 client has rebooted.
1165 A client uses a particular security flavor and principal when perform‐
1167 ing the operations to establish a lease. If two clients happen to
1168 present the same identity string, a server can use their principals to
1169 detect that they are different clients, and prevent one client from
1170 interfering with the other's lease.
1171 The Linux NFS client establishes one lease for each server. Lease man‐
1173 agement operations, such as lease renewal, are not done on behalf of a
1174 particular file, lock, user, or mount point, but on behalf of the whole
1175 client that owns that lease. These operations must use the same secu‐
1176 rity flavor and principal that was used when the lease was established,
1177 even across client reboots.
1178 When Kerberos is configured on a Linux NFS client (i.e., there is a
1180 /etc/krb5.keytab on that client), the client attempts to use a Kerberos
1181 security flavor for its lease management operations. This provides
1182 strong authentication of the client to each server it contacts. By
1183 default, the client uses the host/ or nfs/ service principal in its
1184 /etc/krb5.keytab for this purpose.
1185 If the client has Kerberos configured, but the server does not, or if
1187 the client does not have a keytab or the requisite service principals,
1188 the client uses AUTH_SYS and UID 0 for lease management.
1189 Using non-privileged source ports
1191 NFS clients usually communicate with NFS servers via network sockets.
1192 Each end of a socket is assigned a port value, which is simply a number
1193 between 1 and 65535 that distinguishes socket endpoints at the same IP
1194 address. A socket is uniquely defined by a tuple that includes the
1195 transport protocol (TCP or UDP) and the port values and IP addresses of
1196 both endpoints.
1197 The NFS client can choose any source port value for its sockets, but
1199 usually chooses a privileged port. A privileged port is a port value
1200 less than 1024. Only a process with root privileges may create a
1201 socket with a privileged source port.
1202 The exact range of privileged source ports that can be chosen is set by
1204 a pair of sysctls to avoid choosing a well-known port, such as the port
1205 used by ssh. This means the number of source ports available for the
1206 NFS client, and therefore the number of socket connections that can be
1207 used at the same time, is practically limited to only a few hundred.
1208 As described above, the traditional default NFS authentication scheme,
1210 known as AUTH_SYS, relies on sending local UID and GID numbers to iden‐
1211 tify users making NFS requests. An NFS server assumes that if a con‐
1212 nection comes from a privileged port, the UID and GID numbers in the
1213 NFS requests on this connection have been verified by the client's ker‐
1214 nel or some other local authority. This is an easy system to spoof,
1215 but on a trusted physical network between trusted hosts, it is entirely
1216 adequate.
1217 Roughly speaking, one socket is used for each NFS mount point. If a
1219 client could use non-privileged source ports as well, the number of
1220 sockets allowed, and thus the maximum number of concurrent mount
1221 points, would be much larger.
1222 Using non-privileged source ports may compromise server security some‐
1224 what, since any user on AUTH_SYS mount points can now pretend to be any
1225 other when making NFS requests. Thus NFS servers do not support this
1226 by default. They explicitly allow it usually via an export option.
1227 To retain good security while allowing as many mount points as possi‐
1229 ble, it is best to allow non-privileged client connections only if the
1230 server and client both require strong authentication, such as Kerberos.
1231 Mounting through a firewall
1233 A firewall may reside between an NFS client and server, or the client
1234 or server may block some of its own ports via IP filter rules. It is
1235 still possible to mount an NFS server through a firewall, though some
1236 of the mount(8) command's automatic service endpoint discovery mecha‐
1237 nisms may not work; this requires you to provide specific endpoint
1238 details via NFS mount options.
1239 NFS servers normally run a portmapper or rpcbind daemon to advertise
1241 their service endpoints to clients. Clients use the rpcbind daemon to
1242 determine:
1243 What network port each RPC-based service is using
1245 What transport protocols each RPC-based service supports
1247 The rpcbind daemon uses a well-known port number (111) to help clients
1249 find a service endpoint. Although NFS often uses a standard port num‐
1250 ber (2049), auxiliary services such as the NLM service can choose any
1251 unused port number at random.
1252 Common firewall configurations block the well-known rpcbind port. In
1254 the absense of an rpcbind service, the server administrator fixes the
1255 port number of NFS-related services so that the firewall can allow
1256 access to specific NFS service ports. Client administrators then spec‐
1257 ify the port number for the mountd service via the mount(8) command's
1258 mountport option. It may also be necessary to enforce the use of TCP
1259 or UDP if the firewall blocks one of those transports.
1260 NFS Access Control Lists
1262 Solaris allows NFS version 3 clients direct access to POSIX Access Con‐
1263 trol Lists stored in its local file systems. This proprietary sideband
1264 protocol, known as NFSACL, provides richer access control than mode
1265 bits. Linux implements this protocol for compatibility with the
1266 Solaris NFS implementation. The NFSACL protocol never became a stan‐
1267 dard part of the NFS version 3 specification, however.
1268 The NFS version 4 specification mandates a new version of Access Con‐
1270 trol Lists that are semantically richer than POSIX ACLs. NFS version 4
1271 ACLs are not fully compatible with POSIX ACLs; as such, some transla‐
1272 tion between the two is required in an environment that mixes POSIX
1273 ACLs and NFS version 4.
1274
1276 Generic mount options such as rw and sync can be modified on NFS mount
1277 points using the remount option. See mount(8) for more information on
1278 generic mount options.
1279 With few exceptions, NFS-specific options are not able to be modified
1281 during a remount. The underlying transport or NFS version cannot be
1282 changed by a remount, for example.
1283 Performing a remount on an NFS file system mounted with the noac option
1285 may have unintended consequences. The noac option is a combination of
1286 the generic option sync, and the NFS-specific option actimeo=0.
1287 Unmounting after a remount
1289 For mount points that use NFS versions 2 or 3, the NFS umount subcom‐
1290 mand depends on knowing the original set of mount options used to per‐
1291 form the MNT operation. These options are stored on disk by the NFS
1292 mount subcommand, and can be erased by a remount.
1293 To ensure that the saved mount options are not erased during a remount,
1295 specify either the local mount directory, or the server hostname and
1296 export pathname, but not both, during a remount. For example,
1297 mount -o remount,ro /mnt
1299 merges the mount option ro with the mount options already saved on disk
1301 for the NFS server mounted at /mnt.
1302
1304 /etc/fstab file system table
1305 /etc/nfsmount.conf
1307 Configuration file for NFS mounts
1308
1310 Before 2.4.7, the Linux NFS client did not support NFS over TCP.
1311 Before 2.4.20, the Linux NFS client used a heuristic to determine
1313 whether cached file data was still valid rather than using the standard
1314 close-to-open cache coherency method described above.
1315 Starting with 2.4.22, the Linux NFS client employs a Van Jacobsen-based
1317 RTT estimator to determine retransmit timeout values when using NFS
1318 over UDP.
1319 Before 2.6.0, the Linux NFS client did not support NFS version 4.
1321 Before 2.6.8, the Linux NFS client used only synchronous reads and
1323 writes when the rsize and wsize settings were smaller than the system's
1324 page size.
1325 The Linux client's support for protocol versions depend on whether the
1327 kernel was built with options CONFIG_NFS_V2, CONFIG_NFS_V3, CON‐
1328 FIG_NFS_V4, CONFIG_NFS_V4_1, and CONFIG_NFS_V4_2.
1329
1331 fstab(5), mount(8), umount(8), mount.nfs(5), umount.nfs(5), exports(5),
1332 nfsmount.conf(5), netconfig(5), ipv6(7), nfsd(8), sm-notify(8),
1333 rpc.statd(8), rpc.idmapd(8), rpc.gssd(8), rpc.svcgssd(8), kerberos(1)
1334 RFC 768 for the UDP specification.
1336 RFC 793 for the TCP specification.
1337 RFC 1094 for the NFS version 2 specification.
1338 RFC 1813 for the NFS version 3 specification.
1339 RFC 1832 for the XDR specification.
1340 RFC 1833 for the RPC bind specification.
1341 RFC 2203 for the RPCSEC GSS API protocol specification.
1342 RFC 7530 for the NFS version 4.0 specification.
1343 RFC 5661 for the NFS version 4.1 specification.
1344 RFC 7862 for the NFS version 4.2 specification.
1345 9 October 2012 NFS(5)