1 .TH EXPORTS 5 "4 March 2005" "Linux" "Linux File Formats Manual"
3 exports \- NFS file systems being exported (for Kernel based NFS)
9 serves as the access control list for file systems which may be
10 exported to NFS clients. It is used by
12 to give information to
14 and to the kernel based NFS file server daemon
17 The file format is similar to the SunOS
19 file. Each line contains an export point and a whitespace-separated list
20 of clients allowed to mount the file system at that point. Each listed
21 client may be immediately followed by a parenthesized, comma-separated
22 list of export options for that client. No whitespace is permitted
23 between a client and its option list.
25 Blank lines are ignored. A pound sign ("#") introduces a comment to the
26 end of the line. Entries may be continued across newlines using a
27 backslash. If an export name contains spaces it should be quoted using
28 double quotes. You can also specify spaces or other unusual character in
29 the export name using a backslash followed by the character code as three
32 To apply changes to this file, run exportfs -ra or (on Debian)
33 /etc/init.d/nfs-kernel-server reload.
35 .SS Machine Name Formats
36 NFS clients may be specified in a number of ways:
38 This is the most common format. You may specify a host either by an
39 abbreviated name recognized be the resolver, the fully qualified domain
40 name, or an IP address.
42 NIS netgroups may be given as
44 Only the host part of each
45 netgroup members is consider in checking for membership. Empty host
46 parts or those containing a single dash (\-) are ignored.
48 Machine names may contain the wildcard characters \fI*\fR and \fI?\fR.
49 This can be used to make the \fIexports\fR file more compact; for instance,
50 \fI*.cs.foo.edu\fR matches all hosts in the domain
51 \fIcs.foo.edu\fR. As these characters also match the dots in a domain
52 name, the given pattern will also match all hosts within any subdomain
55 You can also export directories to all hosts on an IP (sub-) network
56 simultaneously. This is done by specifying an IP address and netmask pair
59 where the netmask can be specified in dotted-decimal format, or as a
60 contiguous mask length (for example, either `/255.255.252.0' or `/22' appended
61 to the network base address result in identical subnetworks with 10 bits of
62 host). Wildcard characters generally do not work on IP addresses, though they
63 may work by accident when reverse DNS lookups fail.
66 '''This is a special ``hostname'' that identifies the given directory name
67 '''as the public root directory (see the section on WebNFS in
69 '''for a discussion of WebNFS and the public root handle). When using this
72 '''must be the only entry on this line, and must have no export options
73 '''associated with it. Note that this does
75 '''actually export the named directory; you still have to set the exports
76 '''options in a separate entry.
78 '''The public root path can also be specified by invoking
81 '''.B \-\-public\-root
82 '''option. Multiple specifications of a public root will be ignored.
84 .SS RPCSEC_GSS security
85 To restrict access to an export using rpcsec_gss security, use the special
86 string "gss/krb5" as the client. It is not possible to simultaneously require
87 rpcsec_gss and to make requirements on the IP address of the client.
91 understands the following export options:
94 This option requires that requests originate on an internet port less
95 than IPPORT_RESERVED (1024). This option is on by default. To turn it
100 Allow both read and write requests on this NFS volume. The
101 default is to disallow any request which changes the filesystem.
102 This can also be made explicit by using
107 This option allows the NFS server to violate the NFS protocol and
108 reply to requests before any changes made by that request have been
109 committed to stable storage (e.g. disc drive).
111 Using this option usually improves performance, but at the cost that
112 an unclean server restart (i.e. a crash) can cause data to be lost or
117 Reply to requests only after the changes have been committed to stable
122 In releases of nfs-utils upto and including 1.0.0, this option was the
123 default. In this and future releases,
127 must be explicit requested if needed.
128 To help make system adminstrators aware of this change, 'exportfs'
129 will issue a warning if neither
136 This option has no effect if
138 is also set. The NFS server will normally delay committing a write request
139 to disc slightly if it suspects that another related write request may be in
140 progress or may arrive soon. This allows multiple write requests to
141 be committed to disc with the one operation which can improve
142 performance. If an NFS server received mainly small unrelated
143 requests, this behaviour could actually reduce performance, so
145 is available to turn it off.
146 The default can be explicitly requested with the
150 This option is based on the option of the same name provided in IRIX
151 NFS. Normally, if a server exports two filesystems one of which is
152 mounted on the other, then the client will have to mount both
153 filesystems explicitly to get access to them. If it just mounts the
154 parent, it will see an empty directory at the place where the other
155 filesystem is mounted. That filesystem is "hidden".
159 option on a filesystem causes it not to be hidden, and an
160 appropriately authorised client will be able to move from the parent to
161 that filesystem without noticing the change.
163 However, some NFS clients do not cope well with this situation as, for
164 instance, it is then possible for two files in the one apparent
165 filesystem to have the same inode number.
169 option is currently only effective on
171 exports. It does not work reliably with netgroup, subnet, or wildcard
174 This option can be very useful in some situations, but it should be
175 used with due care, and only after confirming that the client system
176 copes with the situation effectively.
178 The option can be explicitly disabled with
182 This option is similar to
184 but it makes it possible for clients to move from the filesystem marked
185 with crossmnt to exported filesystems mounted on it. Thus when a child
186 filesystem "B" is mounted on a parent "A", setting crossmnt on "A" has
187 the same effect as setting "nohide" on B.
190 This option disables subtree checking, which has mild security
191 implications, but can improve reliability in some circumstances.
193 If a subdirectory of a filesystem is exported, but the whole
194 filesystem isn't then whenever a NFS request arrives, the server must
195 check not only that the accessed file is in the appropriate filesystem
196 (which is easy) but also that it is in the exported tree (which is
197 harder). This check is called the
200 In order to perform this check, the server must include some
201 information about the location of the file in the "filehandle" that is
202 given to the client. This can cause problems with accessing files that
203 are renamed while a client has them open (though in many simple cases
206 subtree checking is also used to make sure that files inside
207 directories to which only root has access can only be accessed if the
208 filesystem is exported with
210 (see below), even if the file itself allows more general access.
212 As a general guide, a home directory filesystem, which is normally
213 exported at the root and may see lots of file renames, should be
214 exported with subtree checking disabled. A filesystem which is mostly
215 readonly, and at least doesn't see many file renames (e.g. /usr or
216 /var) and for which subdirectories may be exported, should probably be
217 exported with subtree checks enabled.
219 The default of having subtree checks enabled, can be explicitly
227 This option (the two names are synonymous) tells the NFS server not to require authentication of
228 locking requests (i.e. requests which use the NLM protocol). Normally
229 the NFS server will require a lock request to hold a credential for a
230 user who has read access to the file. With this flag no access checks
233 Early NFS client implementations did not send credentials with lock
234 requests, and many current NFS clients still exist which are based on
235 the old implementations. Use this flag if you find that you can only
236 lock files which are world readable.
238 The default behaviour of requiring authentication for NLM requests can
239 be explicitly requested with either of the synonymous
245 On some specially patched kernels, and when exporting filesystems that
246 support ACLs, this option tells nfsd not to reveal ACLs to clients, so
247 they will see only a subset of actual permissions on the given file
248 system. This option is safe for filesystems used by NFSv2 clients and
249 old NFSv3 clients that perform access decisions locally. Current
250 NFSv3 clients use the ACCESS RPC to perform all access decisions on
251 the server. Note that the
253 option only has effect on kernels specially patched to support it, and
254 when exporting filesystems with ACL support. The default is to export
255 with ACL support (i.e. by default,
261 '''This makes everything below the directory inaccessible for the named
262 '''client. This is useful when you want to export a directory hierarchy to
263 '''a client, but exclude certain subdirectories. The client's view of a
264 '''directory flagged with noaccess is very limited; it is allowed to read
265 '''its attributes, and lookup `.' and `..'. These are also the only entries
266 '''returned by a readdir.
269 '''Convert absolute symbolic links (where the link contents start with a
270 '''slash) into relative links by prepending the necessary number of ../'s
271 '''to get from the directory containing the link to the root on the
272 '''server. This has subtle, perhaps questionable, semantics when the file
273 '''hierarchy is not mounted at its root.
276 '''Leave all symbolic link as they are. This is the default operation.
282 This option makes it possible to only export a directory if it has
283 successfully been mounted.
284 If no path is given (e.g.
285 .IR mountpoint " or " mp )
286 then the export point must also be a mount point. If it isn't then
287 the export point is not exported. This allows you to be sure that the
288 directory underneath a mountpoint will never be exported by accident
289 if, for example, the filesystem failed to mount due to a disc error.
291 If a path is given (e.g.
292 .IR mountpoint= "/path or " mp= /path)
293 then the nominted path must be a mountpoint for the exportpoint to be
298 This option forces the filesystem identification portion of the file
299 handle and file attributes used on the wire to be
301 instead of a number derived from the major and minor number of the
302 block device on which the filesystem is mounted. Any 32 bit number
303 can be used, but it must be unique amongst all the exported filesystems.
305 This can be useful for NFS failover, to ensure that both servers of
306 the failover pair use the same NFS file handles for the shared filesystem
307 thus avoiding stale file handles after failover.
309 Some Linux filesystems are not mounted on a block device; exporting
310 these via NFS requires the use of the
312 option (although that may still not be enough).
314 The value 0 has a special meaning when use with NFSv4. NFSv4 has a
315 concept of a root of the overall exported filesystem. The export point
316 exported with fsid=0 will be used as this root.
321 bases its access control to files on the server machine on the uid and
322 gid provided in each NFS RPC request. The normal behavior a user would
323 expect is that she can access her files on the server just as she would
324 on a normal file system. This requires that the same uids and gids are
325 used on the client and the server machine. This is not always true, nor
326 is it always desirable.
328 Very often, it is not desirable that the root user on a client machine
329 is also treated as root when accessing files on the NFS server. To this
330 end, uid 0 is normally mapped to a different id: the so-called
333 uid. This mode of operation (called `root squashing') is the default,
334 and can be turned off with
339 '''tries to obtain the anonymous uid and gid by looking up user
341 '''in the password file at startup time. If it isn't found, a uid and gid
343 chooses a uid and gid
344 of 65534 for squashed access. These values can also be overridden by
346 .IR anonuid " and " anongid
349 '''In addition to this,
351 '''lets you specify arbitrary uids and gids that should be mapped to user
353 Finally, you can map all user requests to the
354 anonymous uid by specifying the
355 .IR all_squash " option.
357 '''For the benefit of installations where uids differ between different
360 '''provides several mechanism to dynamically map server uids to client
361 '''uids and vice versa: static mapping files, NIS-based mapping, and
366 '''mapping is enabled with the
368 '''option, and uses the UGID RPC protocol. For this to work, you have to run
371 '''mapping daemon on the client host. It is the least secure of the three methods,
372 '''because by running
374 '''everybody can query the client host for a list of valid user names. You
375 '''can protect yourself by restricting access to
377 '''to valid hosts only. This can be done by entering the list of valid
382 '''file. The service name is
384 '''For a description of the file's syntax, please read
385 '''.IR hosts_access (5).
387 '''Static mapping is enabled by using the
389 '''option, which takes a file name as an argument that describes the mapping.
390 '''NIS-based mapping queries the client's NIS server to obtain a mapping from
391 '''user and group names on the server host to user and group names on the
394 Here's the complete list of mapping options:
397 Map requests from uid/gid 0 to the anonymous uid/gid. Note that this does
398 not apply to any other uids that might be equally sensitive, such as user
402 Turn off root squashing. This option is mainly useful for diskless clients.
404 '''.IR squash_uids " and " squash_gids
405 '''This option specifies a list of uids or gids that should be subject to
406 '''anonymous mapping. A valid list of ids looks like this:
408 '''.IR squash_uids=0-15,20,25-50
410 '''Usually, your squash lists will look a lot simpler.
413 Map all uids and gids to the anonymous user. Useful for NFS-exported
414 public FTP directories, news spool directories, etc. The opposite option
417 which is the default setting.
420 '''This option turns on dynamic uid/gid mapping. Each uid in an NFS request
421 '''will be translated to the equivalent server uid, and each uid in an
422 '''NFS reply will be mapped the other way round. This option requires that
424 '''runs on the client host. The default setting is
425 '''.IR map_identity ,
426 '''which leaves all uids untouched. The normal squash options apply regardless
427 '''of whether dynamic mapping is requested or not.
430 '''This option enables static mapping. It specifies the name of the file
431 '''that describes the uid/gid mapping, e.g.
433 '''.IR map_static=/etc/nfs/foobar.map
435 '''The file's format looks like this
439 '''# Mapping for client foobar:
441 '''uid 0-99 - # squash these
442 '''uid 100-500 1000 # map 100-500 to 1000-1400
443 '''gid 0-49 - # squash these
444 '''gid 50-100 700 # map 50-100 to 700-750
448 '''This option enables NIS-based uid/gid mapping. For instance, when
449 '''the server encounters the uid 123 on the server, it will obtain the
450 '''login name associated with it, and contact the NFS client's NIS server
451 '''to obtain the uid the client associates with the name.
453 '''In order to do this, the NFS server must know the client's NIS domain.
454 '''This is specified as an argument to the
458 '''.I map_nis=foo.com
460 '''Note that it may not be sufficient to simply specify the NIS domain
461 '''here; you may have to take additional actions before
463 '''is actually able to contact the server. If your distribution uses
464 '''the NYS library, you can specify one or more NIS servers for the
465 '''client's domain in
466 '''.IR /etc/yp.conf .
467 '''If you are using a different NIS library, you may have to obtain a
470 '''daemon that can be configured via
473 .IR anonuid " and " anongid
474 These options explicitly set the uid and gid of the anonymous account.
475 This option is primarily useful for PC/NFS clients, where you might want
476 all requests appear to be from one user. As an example, consider the
479 in the example section below, which maps all requests to uid 150 (which
480 is supposedly that of user joe).
486 # sample /etc/exports file
487 / master(rw) trusty(rw,no_root_squash)
488 /projects proj*.local.domain(rw)
489 /usr *.local.domain(ro) @trusted(rw)
490 /home/joe pc001(rw,all_squash,anonuid=150,anongid=100)
491 /pub (ro,insecure,all_squash)
492 '''/pub/private (noaccess)
495 The first line exports the entire filesystem to machines master and trusty.
496 In addition to write access, all uid squashing is turned off for host
497 trusty. The second and third entry show examples for wildcard hostnames
498 and netgroups (this is the entry `@trusted'). The fourth line shows the
499 entry for the PC/NFS client discussed above. Line 5 exports the
500 public FTP directory to every host in the world, executing all requests
501 under the nobody account. The
503 option in this entry also allows clients with NFS implementations that
504 don't use a reserved port for NFS.
505 ''' The last line denies all NFS clients
506 '''access to the private directory.
508 '''Unlike other NFS server implementations, this
510 '''allows you to export both a directory and a subdirectory thereof to
511 '''the same host, for instance
512 '''.IR /usr " and " /usr/X11R6 .
513 '''In this case, the mount options of the most specific entry apply. For
514 '''instance, when a user on the client host accesses a file in
516 '''the mount options given in the
518 '''entry apply. This is also true when the latter is a wildcard or netgroup
529 '''An error parsing the file is reported using syslogd(8) as level NOTICE from
530 '''a DAEMON whenever nfsd(8) or mountd(8) is started up. Any unknown
531 '''host is reported at that time, but often not all hosts are not yet known
532 '''to named(8) at boot time, thus as hosts are found they are reported
533 '''with the same syslogd(8) parameters.