NAME

mount_nfs — mount NFS file systems

SYNOPSIS

DESCRIPTION

The mount_nfs command calls the mount(2) system call to prepare and graft a remote NFS file system ( server:/path ) on to the file system tree at the point directory.

This command is expected to be executed by the mount(8) command. Direct use of mount_nfs to mount NFS file systems is strongly discouraged because there is little practical benefit of using it instead of mount(8).

For NFS versions that use a separate mount protocol, mount_nfs implements the mount protocol as described in RFC 1094, Appendix A and NFS: Network File System Version 3 Protocol Specification, RFC 1813, Appendix I.

By default, mount_nfs will attempt the mount twice before exiting with an error. If the -o bg option is given, it will attempt the mount once and then background itself to continue trying another 10,000 times (pausing for one minute between attempts). The option -o retrycnt=⟨num⟩ can be used if a different retry behavior is desired for a mount.

If the server becomes unresponsive while an NFS file system is mounted, any new or outstanding file operations on that file system will hang uninterruptibly until the server comes back (or that NFS file system is forcibly unmounted). To modify this default behaviour, see the -o intr and -o soft mount options.

Mount options are specified with a -o flag followed by a comma separated string of options. See the mount(8) man page for possible options and their meanings. The following NFS-specific options are also available:

bg

Retry mount in background. If an initial attempt to contact the server fails, fork off a child to keep trying the mount in the background. Useful for startup scripts where the file system mount is not critical to multiuser operation.

retrycnt=⟨num⟩

Set the retry count for doing the mount to the specified value. The default is 1 for foreground mounts and 10,000 for background mounts. Setting retrycnt to 0, in addition to only trying to establish connection once, will case nfs client code to use quick time out value (default 8s) instead of regular mount time out (default 30s) value while establishing initial connection.

udp

Use UDP transport protocol.

tcp

Use the TCP transport protocol instead of UDP. The default is to try TCP first, then fall back to UDP if the server doesn't support TCP.

inet

Use only IPv4 addresses.

inet6

Use only IPv6 addresses.

proto=⟨netid⟩

Use the transport protocol and address family as specified by the given ONC RPC Netid (RFC 5665). Valid netid values are: tcp (TCP over IPv4), udp (UDP over IPv4), tcp6 (TCP over IPv6), and udp6 (UDP over IPv6). Note that this option differs from the separate tcp and udp options described above in that each netid value specifies both a transport protocol and address family (IP version).

mntudp

Force the mount protocol to use UDP transport, even for TCP NFS mounts. (Necessary for some old BSD servers.)

mountport=⟨port-number⟩

Connect to the NFS server's mount daemon using the given port number.

port=⟨port-number⟩

Connect to an NFS server at the given port number.

noconn

Do not connect UDP sockets. For UDP mount points, do not do a connect(2). This must be used for servers that do not reply to requests from the standard NFS port number 2049. It may also be required for servers with more than one IP address if replies come from an address other than the one specified in the requests.

resvport

Use a reserved socket port number. This is useful for mounting servers that require clients to use a reserved port number on the mistaken belief that this makes NFS more secure. (For the rare case where the client has a trusted root account but untrustworthy users and the network cables are in secure areas this does help, but for normal desktop clients this does not apply.) root permission is required to mount using resvport mount option.

intr

Make the mount interruptible, which implies that file system calls that are delayed due to an unresponsive server will fail with EINTR when a termination signal is posted for the process.

soft

Make the mount soft, which means that file system calls will fail after retrans round trip timeout intervals. Note: mounts which are both soft and read-only will also have the locallocks mount option enabled by default - unless explicitly overridden with a lock option (for example, nolocks or nolocallocks ).

vers=⟨num[.num]⟩[-⟨num[.num]⟩]

 

nfsvers=⟨num[.num]⟩[-⟨num[.num]⟩]

NFS protocol version number - 2 for NFSv2, 3 for NFSv3 and 4 for NFSv4. The default is to try version 3 first, and fall back to version 2 if the mount fails. A range of versions can be specified by including a dash and another version with no spaces between versions and the dash. In that case the highest version is tried first and if not successful fall back to each version down to the lowest version specified until the mount succeeds or the lowest version fails. Note minor versions may be specified for versions greater than or equal to four by appending a dot and then the minor version number. Currently NFSv4 is the highest supported version with a minor version of zero. If no minor version is specified, zero is assumed. Specifying a non supported version or minor version will print a warning and ignore the vers or nfsvers option. Versions 2 or 3 do not support minor versioning so minor versions greater than zero are treated as above.

nfsv2

 

nfsv3

 

nfsv4

Deprecated. Use -o vers=⟨num⟩ to specify NFS protocol version.

sec=⟨mechanism⟩

Force a specific security mechanism to be used for the mount, where mechanism is one of: krb5p, krb5i, krb5, or sys. When this option is not given the security mechanism will be negotiated transparently with the remote server.

etype=⟨encryption-type⟩

Use the specified encryption type for the mount, where encryption-type is one of: des3, des3-cbc-sha1, des3-cbc-sha1-kd, aes128, aes128-cts-hmac-sha1, aes128-cts-hmac-sha1-96, aes256-cts-hmac-sha1, or aes256-cts-hmac-sha1-96.

realm=⟨realm⟩

Use the default credential for realm or security domain. For Kerberos realms are usually uppercase. If the realm specified does not begin with an “@”, an “@” sign will be prepended to it. Note specifying the realm is typically used for automounter maps when clients may have multiple credential caches, and tells the client what cache to use on the mount.

principal=⟨principal⟩

Use the specified principal for acquiring credentials for the mount. That principal will be used for all accesses by the mounting credential on the mounted file system. Note specifying a principal is useful for user initiated command line mounts, where the user knows the particular credential to use.

sprincipal=⟨server-principal⟩

Use the specified server-principal for establishing credentials for the mount. That server principal will be use for all mount access. If no server principal is specified, then the GSS_C_NT_HOSTBASED_SERVICE nfs@server is used, where server is taken from the mount argument server:/path. Note its rare to use this option.

rsize=⟨readsize⟩

Set the read data size to the specified value. The default is 8192 for UDP mounts and 32768 for TCP mounts. It should normally be a power of 2 greater than or equal to 1024. Values greater than 4096 should be multiples of 4096. It may need to be lowered for UDP mounts when the “fragments dropped due to timeout” value is getting large while actively using a mount point. (Use netstat(1) with the -s option to see what the “fragments dropped due to timeout” value is.)

wsize=⟨writesize⟩

Set the write data size to the specified value. Ditto the comments w.r.t. the rsize option, but using the “fragments dropped due to timeout” value on the server instead of the client. Note that both the rsize and wsize options should only be used as a last ditch effort at improving performance when mounting servers that do not support TCP mounts.

rwsize=⟨size⟩

Set both the read data size and write data size to the specified value.

dsize=⟨readdirsize⟩

Set the directory read size to the specified value. The value should normally be a multiple of DIRBLKSIZ that is <= the read size for the mount. The default is 8192 for UDP mounts and 32768 for TCP mounts.

readahead=⟨maxreadahead⟩

Set the maximum read-ahead count to the specified value. The default is 16. This may be in the range of 0 - 128, and determines how many Read RPCs will be read ahead when a large file is being read sequentially. Trying larger values for this is suggested for mounts with a large bandwidth * delay product.

rdirplus / nordirplus

Used with NFS v3/v4 to specify that directory read operations should retrieve additional information about each entry (e.g. use the NFSv3 ReaddirPlus RPC). This option typically reduces RPC traffic for cases such as directory listings that use or display basic attributes (e.g. “ls -F” and “find . -type f” ). Note that the long directory listing format case (i.e. “ls -l” ) may not be helped much when the file system does not natively support extended attributes. Older implementations tended to flood the vnode and name caches with prefetched entries which may not be referenced. The current implementation avoids creating those entries until they are referenced. Try this option and see whether performance improves or degrades. Probably most useful for client to server network interconnects with a large bandwidth times delay product. Default value is rdirplus.

acregmin=⟨seconds⟩

 

acregmax=⟨seconds⟩

 

acdirmin=⟨seconds⟩

 

acdirmax=⟨seconds⟩

 

acrootdirmin=⟨seconds⟩

 

acrootdirmax=⟨seconds⟩

These options set the minimum and maximum attribute cache timeouts for directories, the root directory and "regular" (non-directory) files. The default minimum is 5 seconds and the default maximum is 60 seconds. If not specified, the values of acrootdirmin and acrootdirmax will be assigned to those of acdirmin and acdirmax respectively, to maintain backward compatibility. Setting all the minimum and maximum to zero will disable attribute caching. The algorithm to calculate the timeout is based on the age of the file or directory. The older it is, the longer the attribute cache is considered valid, subject to the limits above. Note that the effectiveness of this algorithm depends on how well the clocks on the client and server are synchronized.

actimeo=⟨seconds⟩

Set all attribute cache timeouts to the same value.

noac

Disable attribute caching. Equivalent to setting actimeo to 0.

nonegnamecache

Disable negative name caching.

accesscache=⟨size⟩

Sets the number of users that can store their access information in the access(2) cache. The cache operates in a First-In-First-Out (FIFO) order, retaining the most recently accessed user information. When the cache reaches its maximum size, older information is evicted. The default cache size is 3, allowing three users to store their access information in the cache simultaneously.

locallocks

For NFSv2/v3 mounts, perform all file locking operations locally on the NFS client (in the VFS layer) instead of on the NFS server. This option can provide file locking support on an NFS file system for which the server does not support file locking. However, because the file locking is only performed on the client, the NFS server and other NFS clients will have no knowledge of the locks. Note: mounts which are both soft and read-only will also have the locallocks mount option enabled by default - unless explicitly overridden with a lock option (for example, nolocks or nolocallocks ).

nolocks

 

nolockd

 

nolock

 

nonlm

For NFSv2/v3 mounts, do not support NFS file locking operations. Any attempt to perform file locking operations on this mount will return the error ENOTSUP regardless of whether or not the NFS server supports NFS file locking.

noquota

Do not support file system quota operations that would normally be serviced by using the RQUOTA protocol. Any attempt to perform quota operations on this mount will return the error ENOTSUP regardless of whether or not the NFS server supports the RQUOTA service.

maxgroups=⟨num⟩

Set the maximum size of the group list for the credentials to the specified value. This should be used for mounts on old servers that cannot handle a group list size of 16, as specified in RFC 1057. Try 8, if users in a lot of groups cannot get a response from the mount point.

dumbtimer

Turn off the dynamic retransmit timeout estimator. This may be useful for UDP mounts that exhibit high retry rates, since it is possible that the dynamically estimated timeout interval is too short.

timeo=⟨timeout⟩

Set the initial retransmit timeout to the specified value (in tenths of a second). The default is 1 second. May be useful for fine tuning UDP mounts over internetworks with high packet loss rates or an overloaded server. Try increasing the interval if nfsstat(1) shows high retransmit rates while the file system is active or reducing the value if there is a low retransmit rate but long response delay observed. (Normally, the dumbtimer option should be specified when using this option to manually tune the timeout interval.)

retrans=⟨count⟩

Set the retransmit timeout count for soft mounts to the specified value. The default value is 10.

deadtimeout=⟨timeout⟩

If the mount is still unresponsive timeout seconds after it is initially reported unresponsive, then mark the mount as dead so that it will be forcibly unmounted. Note: mounts which are both soft and read-only will also have the deadtimeout mount option set to 60 seconds. This can be explicitly overridden by setting deadtimeout=0.

mutejukebox

When NFS requests repeatedly get jukebox errors (NFS3ERR_JUKEBOX, NFS4ERR_DELAY) from the server the NFS file system is reported as being unresponsive. Use of this option will prevent the file system from being included in the list of unresponsive file systems that would be included in a dialog presented to the user. This option may be useful when a file system is expected to get such errors during normal operation. For example, when it's backed by a hierarchical storage management system.

async

Assume that unstable write requests have actually been committed to stable storage on the server, and thus will not require resending in the event that the server crashes. Use of this option may improve performance but only at the risk of data loss if the server crashes. Note: this mount option will only be honored if the nfs.client.allow_async option in nfs.conf(5) is also enabled.

sync

Perform I/O requests (specifically, write requests) synchronously. The operation will not return until a response is received from the server. (The default, nosync, behavior is to return once the I/O has been queued up.)

nocallback

For NFSv4 mounts, don't support callback requests from the server. This should effectively disable features that require callback requests such as delegations.

nonamedattr

For NFSv4 mounts, don't support named attributes even if the server does. This is the default.

namedattr

For NFSv4 mounts, if the server appears to support named attributes, they will be used to store extended attributes and named streams (e.g. FinderInfo and resource forks).

noacl

For NFSv4 mounts, don't support ACLs even if the server does. ACLs are currently disabled by default to avoid issues with the way ACLs and modes are handled differently on other operating systems. This may be overriden by specifying the acl option.

aclonly

For NFSv4 mounts, only support ACLs; do not support the mode attribute. (Any mode attribute values returned will have all permission bits set - regardless of the value of any ACL or access mode stored in the file system.) This option overrides the noacl option.

nfc

Convert name strings to Unicode Normalization Form C (NFC) when sending them to the NFS server. This option may be used to improve interoperability with NFS clients and servers that typically use names in the NFC form.

nfs.conf(5) can be used to configure some NFS client options. In particular, nfs.client.mount.options can be used to specify default mount options. This can be useful in situations where it is not easy to configure the command-line options. Some NFS client options in nfs.conf(5) correspond to kernel configuration values which will get set by mount_nfs when performing a mount. To update these values without performing a mount, use the command: mount_nfs configupdate.

COMPATIBILITY

The following mount_nfs command line flags have equivalent -o option forms (shown in parentheses) and their use is strongly discouraged. These command line flags are deprecated and the -o option forms should be used instead.

-2 (vers=2), -3 (vers=3), -4 (vers=4), -L (nolocks), -P (resvport), -T (tcp), -U (mntudp), -b (bg), -c (noconn), -d (dumbtimer), -i (intr), -l (rdirplus), -m (nordirplus), -s (soft), -I readdirsize (dsize=#), -R retrycnt (retrycnt=#), -a maxreadahead (readahead=#), -g maxgroups (maxgroups=#), -r readsize (rsize=#), -t timeout (timeo=#), -w writesize (wsize=#), -x retrans (retrans=#).

EXAMPLES

The simplest way to invoke mount_nfs is with a command like:

or:

PERFORMANCE

As can be derived from the comments accompanying the options, performance tuning of NFS can be a non-trivial task. Here are some common points to watch:

• Use of the sync option will probably have a detrimental affect on performance. Its use is discouraged as it provides little benefit.
• Use of the async option may improve performance, but only at the risk of losing data if the server crashes because the client will not be making sure that all data is committed to stable storage on the server.
• Increasing the read and write size with the rsize and wsize options respectively will increase throughput if the network interface can handle the larger packet sizes.

  The default read and write sizes are 8K when using UDP, and 32K when using TCP. Values over 16K are only supported for TCP, where 2M is the maximum.

  Any value over 32K is unlikely to get you more performance, unless you have a very fast network.

• If the network interface cannot handle larger packet sizes or a long train of back to back packets, you may see low performance figures or even temporary hangups during NFS activity.

  This can especially happen with lossy network connections (e.g. wireless networks) which can lead to a lot of dropped packets.

  In this case, decreasing the read and write size, using TCP, or a combination of both will usually lead to better throughput.

• For connections that are not on the same LAN, and/or may experience packet loss, using TCP is strongly recommended.

ERRORS

Some common problems with mount_nfs can be difficult for first time users to understand.

This message means that the remote host is either not exporting the file system you requested or is not exporting it to your host. If you believe the remote host is indeed exporting a file system to you, make sure the exports(5) file is exporting the proper directories. The program showmount(8) can be used to see a server's exports list. The command “showmount -e remotehostname” will display what file systems the remote host is exporting.

A common mistake is that mountd(8) will not export a file system with the -alldirs option, unless it is a mount point on the exporting host. It is not possible to remotely mount a subdirectory of an exported mount, unless it is exported with the -alldirs option.

The following error:

means that the remote host is not running nfsd(8). or mountd(8). The program rpcinfo(8) can be used to determine if the remote host is running nfsd and mountd by issuing the command:

If the remote host is running nfsd, mountd, rpc.statd, and rpc.lockd it would display:

The following error:

indicates that mount_nfs could not resolve the name of the remote host.

The following error:

In case the remote host is not running rpc.statd and the client is mounting the NFS directory with remote locks (enabled by default), mount_nfs should fail with EPROGUNAVAIL. rpc.statd status can be verified using rpcinfo (see above).

The following warning:

indicates that provided wsize is much greater than rsize. which could end up with a different amount of readahead RPCs compared to the amount requested by the user.

SEE ALSO

mount(2), unmount(2), mount(8), umount(8), nfsstat(1), netstat(1), rpcinfo(8), showmount(8), automount(8), nfs.conf(5), nfs(5)

CAVEATS

An NFS server shouldn't loopback-mount its own exported file systems because it's fundamentally prone to deadlock.