|File Formats Manual
exports - NFS server export table
The file /etc/exports contains a table of local physical file systems on an NFS server that are accessible to NFS clients. The contents of the file are maintained by the server's system administrator.
The file format is similar to the SunOS exports file. Each line contains an export point and a whitespace-separated list of clients allowed to mount the file system at that point. Each listed client may be immediately followed by a parenthesized, comma-separated list of export options for that client. No whitespace is permitted between a client and its option list.
Also, each line may have one or more specifications for default options after the path name, in the form of a dash ("-") followed by an option list. The option list is used for all subsequent exports on that line only.
Blank lines are ignored. A pound sign ("#") introduces a comment to the end of the line. Entries may be continued across newlines using a backslash. If an export name contains spaces it should be quoted using double quotes. You can also specify spaces or other unusual character in the export name using a backslash followed by the character code as three octal digits.
To apply changes to this file, run exportfs -ra or restart the NFS server.
Machine Name Formats¶
NFS clients may be specified in a number of ways:
- single host
- You may specify a host either by an abbreviated name recognized be the resolver, the fully qualified domain name, an IPv4 address, or an IPv6 address. IPv6 addresses must not be inside square brackets in /etc/exports lest they be confused with character-class wildcard matches.
- IP networks
- You can also export directories to all hosts on an IP (sub-) network simultaneously. This is done by specifying an IP address and netmask pair as address/netmask where the netmask can be specified in dotted-decimal format, or as a contiguous mask length. For example, either `/255.255.252.0' or `/22' appended to the network base IPv4 address results in identical subnetworks with 10 bits of host. IPv6 addresses must use a contiguous mask length and must not be inside square brackets to avoid confusion with character-class wildcards. Wildcard characters generally do not work on IP addresses, though they may work by accident when reverse DNS lookups fail.
- Machine names may contain the wildcard characters * and ?, or may contain character class lists within [square brackets]. This can be used to make the exports file more compact; for instance, *.cs.foo.edu matches all hosts in the domain cs.foo.edu. As these characters also match the dots in a domain name, the given pattern will also match all hosts within any subdomain of cs.foo.edu.
- NIS netgroups may be given as @group. Only the host part of each netgroup members is consider in checking for membership. Empty host parts or those containing a single dash (-) are ignored.
- This is specified by a single * character (not to be confused with the wildcard entry above) and will match all clients.
If a client matches more than one of the specifications above, then the first match from the above list order takes precedence - regardless of the order they appear on the export line. However, if a client matches more than one of the same type of specification (e.g. two netgroups), then the first match from the order they appear on the export line takes precedence.
You may use the special strings "gss/krb5", "gss/krb5i", or "gss/krb5p" to restrict access to clients using rpcsec_gss security. However, this syntax is deprecated; on linux kernels since 2.6.23, you should instead use the "sec=" export option:
- The sec= option, followed by a colon-delimited list of security flavors, restricts the export to clients using those flavors. Available security flavors include sys (the default--no cryptographic security), krb5 (authentication only), krb5i (integrity protection), and krb5p (privacy protection). For the purposes of security flavor negotiation, order counts: preferred flavors should be listed first. The order of the sec= option with respect to the other options does not matter, unless you want some options to be enforced differently depending on flavor. In that case you may include multiple sec= options, and following options will be enforced only for access using flavors listed in the immediately preceding sec= option. The only options that are permitted to vary in this way are ro, rw, no_root_squash, root_squash, and all_squash.
exportfs understands the following export options:
- This option requires that requests not using gss originate on an Internet port less than IPPORT_RESERVED (1024). This option is on by default. To turn it off, specify insecure. (NOTE: older kernels (before upstream kernel version 4.17) enforced this requirement on gss requests as well.)
- Allow both read and write requests on this NFS volume. The default is to disallow any request which changes the filesystem. This can also be made explicit by using the ro option.
- This option allows the NFS server to violate the NFS protocol and reply to
requests before any changes made by that request have been committed to
stable storage (e.g. disc drive).
Using this option usually improves performance, but at the cost that an unclean server restart (i.e. a crash) can cause data to be lost or corrupted.
- Reply to requests only after the changes have been committed to stable
storage (see async above).
In releases of nfs-utils up to and including 1.0.0, the async option was the default. In all releases after 1.0.0, sync is the default, and async must be explicitly requested if needed.
- This option has no effect if async is also set. The NFS server will normally delay committing a write request to disc slightly if it suspects that another related write request may be in progress or may arrive soon. This allows multiple write requests to be committed to disc with the one operation which can improve performance. If an NFS server received mainly small unrelated requests, this behaviour could actually reduce performance, so no_wdelay is available to turn it off. The default can be explicitly requested with the wdelay option.
- This option is based on the option of the same name provided in IRIX NFS.
Normally, if a server exports two filesystems one of which is mounted on
the other, then the client will have to mount both filesystems explicitly
to get access to them. If it just mounts the parent, it will see an empty
directory at the place where the other filesystem is mounted. That
filesystem is "hidden".
Setting the nohide option on a filesystem causes it not to be hidden, and an appropriately authorised client will be able to move from the parent to that filesystem without noticing the change.
However, some NFS clients do not cope well with this situation as, for instance, it is then possible for two files in the one apparent filesystem to have the same inode number.
The nohide option is currently only effective on single host exports. It does not work reliably with netgroup, subnet, or wildcard exports.
This option can be very useful in some situations, but it should be used with due care, and only after confirming that the client system copes with the situation effectively.
The option can be explicitly disabled for NFSv2 and NFSv3 with hide.
This option is not relevant when NFSv4 is use. NFSv4 never hides subordinate filesystems. Any filesystem that is exported will be visible where expected when using NFSv4.
- This option is similar to nohide but it makes it possible for
clients to access all filesystems mounted on a filesystem marked with
crossmnt. Thus when a child filesystem "B" is mounted on
a parent "A", setting crossmnt on "A" has a similar
effect to setting "nohide" on B.
With nohide the child filesystem needs to be explicitly exported. With crossmnt it need not. If a child of a crossmnt file is not explicitly exported, then it will be implicitly exported with the same export options as the parent, except for fsid=. This makes it impossible to not export a child of a crossmnt filesystem. If some but not all subordinate filesystems of a parent are to be exported, then they must be explicitly exported and the parent should not have crossmnt set.
The nocrossmnt option can explictly disable crossmnt if it was previously set. This is rarely useful.
- This option disables subtree checking, which has mild security
implications, but can improve reliability in some circumstances.
If a subdirectory of a filesystem is exported, but the whole filesystem isn't then whenever a NFS request arrives, the server must check not only that the accessed file is in the appropriate filesystem (which is easy) but also that it is in the exported tree (which is harder). This check is called the subtree_check.
In order to perform this check, the server must include some information about the location of the file in the "filehandle" that is given to the client. This can cause problems with accessing files that are renamed while a client has them open (though in many simple cases it will still work).
subtree checking is also used to make sure that files inside directories to which only root has access can only be accessed if the filesystem is exported with no_root_squash (see below), even if the file itself allows more general access.
As a general guide, a home directory filesystem, which is normally exported at the root and may see lots of file renames, should be exported with subtree checking disabled. A filesystem which is mostly readonly, and at least doesn't see many file renames (e.g. /usr or /var) and for which subdirectories may be exported, should probably be exported with subtree checks enabled.
The default of having subtree checks enabled, can be explicitly requested with subtree_check.
From release 1.1.0 of nfs-utils onwards, the default will be no_subtree_check as subtree_checking tends to cause more problems than it is worth. If you genuinely require subtree checking, you should explicitly put that option in the exports file. If you put neither option, exportfs will warn you that the change is pending.
- This option (the two names are synonymous) tells the NFS server not to
require authentication of locking requests (i.e. requests which use the
NLM protocol). Normally the NFS server will require a lock request to hold
a credential for a user who has read access to the file. With this flag no
access checks will be performed.
Early NFS client implementations did not send credentials with lock requests, and many current NFS clients still exist which are based on the old implementations. Use this flag if you find that you can only lock files which are world readable.
The default behaviour of requiring authentication for NLM requests can be explicitly requested with either of the synonymous auth_nlm, or secure_locks.
- This option makes it possible to only export a directory if it has
successfully been mounted. If no path is given (e.g. mountpoint or
mp) then the export point must also be a mount point. If it isn't
then the export point is not exported. This allows you to be sure that the
directory underneath a mountpoint will never be exported by accident if,
for example, the filesystem failed to mount due to a disc error.
If a path is given (e.g. mountpoint=/path or mp=/path) then the nominated path must be a mountpoint for the exportpoint to be exported.
- NFS needs to be able to identify each filesystem that it exports. Normally
it will use a UUID for the filesystem (if the filesystem has such a thing)
or the device number of the device holding the filesystem (if the
filesystem is stored on the device).
As not all filesystems are stored on devices, and not all filesystems have UUIDs, it is sometimes necessary to explicitly tell NFS how to identify a filesystem. This is done with the fsid= option.
For NFSv4, there is a distinguished filesystem which is the root of all exported filesystem. This is specified with fsid=root or fsid=0 both of which mean exactly the same thing.
Other filesystems can be identified with a small integer, or a UUID which should contain 32 hex digits and arbitrary punctuation.
Linux kernels version 2.6.20 and earlier do not understand the UUID setting so a small integer must be used if an fsid option needs to be set for such kernels. Setting both a small number and a UUID is supported so the same configuration can be made to work on old and new kernels alike.
- This option will disable READDIRPLUS request handling. When set, READDIRPLUS requests from NFS clients return NFS3ERR_NOTSUPP, and clients fall back on READDIR. This option affects only NFSv3 clients.
- A client referencing the export point will be directed to choose from the given list an alternative location for the filesystem. (Note that the server must have a mountpoint here, though a different filesystem is not required; so, for example, mount --bind /path /path is sufficient.)
- If the client asks for alternative locations for the export point, it will be given this list of alternatives. (Note that actual replication of the filesystem must be handled elsewhere.)
- This option enables the use of the pNFS extension if the protocol level is NFSv4.1 or higher, and the filesystem supports pNFS exports. With pNFS clients can bypass the server and perform I/O directly to storage devices. The default can be explicitly requested with the no_pnfs option.
- With this option set, clients using NFSv4.2 or higher will be able to set and retrieve security labels (such as those used by SELinux). This will only work if all clients use a consistent security policy. Note that early kernels did not support this export option, and instead enabled security labels by default.
User ID Mapping¶
nfsd bases its access control to files on the server machine on the uid and gid provided in each NFS RPC request. The normal behavior a user would expect is that she can access her files on the server just as she would on a normal file system. This requires that the same uids and gids are used on the client and the server machine. This is not always true, nor is it always desirable.
Very often, it is not desirable that the root user on a client machine is also treated as root when accessing files on the NFS server. To this end, uid 0 is normally mapped to a different id: the so-called anonymous or nobody uid. This mode of operation (called `root squashing') is the default, and can be turned off with no_root_squash.
By default, exportfs chooses a uid and gid of 65534 for squashed access. These values can also be overridden by the anonuid and anongid options. Finally, you can map all user requests to the anonymous uid by specifying the all_squash option.
Here's the complete list of mapping options:
- Map requests from uid/gid 0 to the anonymous uid/gid. Note that this does not apply to any other uids or gids that might be equally sensitive, such as user bin or group staff.
- Turn off root squashing. This option is mainly useful for diskless clients.
- Map all uids and gids to the anonymous user. Useful for NFS-exported public FTP directories, news spool directories, etc. The opposite option is no_all_squash, which is the default setting.
- anonuid and anongid
- These options explicitly set the uid and gid of the anonymous account. This option is primarily useful for PC/NFS clients, where you might want all requests appear to be from one user. As an example, consider the export entry for /home/joe in the example section below, which maps all requests to uid 150 (which is supposedly that of user joe).
Normally you should only export only the root of a filesystem. The NFS server will also allow you to export a subdirectory of a filesystem, however, this has drawbacks:
First, it may be possible for a malicious user to access files on the filesystem outside of the exported subdirectory, by guessing filehandles for those other files. The only way to prevent this is by using the no_subtree_check option, which can cause other problems.
Second, export options may not be enforced in the way that you would expect. For example, the security_label option will not work on subdirectory exports, and if nested subdirectory exports change the security_label or sec= options, NFSv4 clients will normally see only the options on the parent export. Also, where security options differ, a malicious client may use filehandle-guessing attacks to access the files from one subdirectory using the options from another.
Extra Export Tables¶
After reading /etc/exports exportfs reads files in the /etc/exports.d directory as extra export tables. Only files ending in .exports are considered. Files beginning with a dot are ignored. The format for extra export tables is the same as /etc/exports
# sample /etc/exports file / master(rw) trusty(rw,no_root_squash) /projects proj*.local.domain(rw) /usr *.local.domain(ro) @trusted(rw) /home/joe pc001(rw,all_squash,anonuid=150,anongid=100) /pub *(ro,insecure,all_squash) /srv/www -sync,rw server @trusted @external(ro) /foo 2001:db8:9:e54::/64(rw) 192.0.2.0/24(rw) /build buildhost[0-9].local.domain(rw)
The first line exports the entire filesystem to machines master and trusty. In addition to write access, all uid squashing is turned off for host trusty. The second and third entry show examples for wildcard hostnames and netgroups (this is the entry `@trusted'). The fourth line shows the entry for the PC/NFS client discussed above. Line 5 exports the public FTP directory to every host in the world, executing all requests under the nobody account. The insecure option in this entry also allows clients with NFS implementations that don't use a reserved port for NFS. The sixth line exports a directory read-write to the machine 'server' as well as the `@trusted' netgroup, and read-only to netgroup `@external', all three mounts with the `sync' option enabled. The seventh line exports a directory to both an IPv6 and an IPv4 subnet. The eighth line demonstrates a character class wildcard match.
|31 December 2009