Copyright (c) 2006-2010 Apple Inc. All rights reserved.

@APPLE_LICENSE_HEADER_START@

This file contains Original Code and/or Modifications of Original Code
as defined in and that are subject to the Apple Public Source License
Version 2.0 (the 'License'). You may not use this file except in
compliance with the License. Please obtain a copy of the License at
http://www.opensource.apple.com/apsl/ and read it before using this
file.

The Original Code and all software distributed under the License are
distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
Please see the License for the specific language governing rights and
limitations under the License.

@APPLE_LICENSE_HEADER_END@

.Dd July 21, 2009 .Dt NFS.CONF 5 .Os .Sh NAME .Nm nfs.conf .Nd The configuration file for .Tn NFS .Sh SYNOPSIS .Nm .Sh DESCRIPTION The .Nm file contains options used to specify .Tn NFS server and client configuration and tuning.

p Each line contains an option field and a value field, separated by an equals character '='. For example: d -literal -offset indent some.nfs.option = value .Ed

p Each line specifies a single option/value pair. Whitespace can be used freely before and after fields. A hash character '#' begins a comment that extends to the end of the line. Lines containing only whitespace or comments are ignored. The file format is intended to be easily read using functions like .Xr fparseln 3 .

p Values are typically integers. For options that enable/disable functionality the value zero (0) indicates the option is off/disabled and the value one (1) (or any other non-zero value) indicates the option is on/enabled.

p The options are: l -tag -width -indent t Cm nfs.client.access_cache_timeout How long (in seconds) that .Xr access 2 information is cached. The default is 60 seconds. t Cm nfs.client.access_for_getattr This option specifies if .Xr access 2 information should be opportunistically fetched every time attributes are fetched. A GETATTR request will only return attributes, but since ACCESS requests usually also return current attributes, a single ACCESS request may be used to fetch both .Xr access 2 information and attributes. Enabling this may improve performance, but only if the ACCESS check on the server is inexpensive. This may not be the case with many modern operating systems. The default is 0 (off). t Cm nfs.client.allow_async Allow the use of the -o .Cm async mount option. This option must be enabled in order for the .Cm async mount option to be honored because (accidental) use of the .Cm async mount option may result in data loss if the server crashes. The default is 0 (off). t Cm nfs.client.callback_port This option can be used to specify a port that the NFSv4 callback RPC service should be available on. The default value is unspecified, which means that any available port will be used. t Cm nfs.client.initialdowndelay When an NFS server is not responding, this option specifies how long to wait (in seconds) before the initial notification is posted. The default is 12 seconds. t Cm nfs.client.iosize This option specifies what size (in bytes) the NFS client reports for the recommended I/O request size returned in .Xr stat 2 and .Xr statfs 2 calls. The default value is 1048576 bytes. t Cm nfs.client.mount.options Mount options to be used for NFS file systems mounted via .Xr mount 8 / .Xr mount_nfs 8 . The value is in the same format as the argument for the .Fl o mount option (a comma-separated string of options like: option1,option2=val,option3). The default value is empty. These options are processed by .Xr mount_nfs 8 prior to processing any other command-line options. Therefore, mount options set in the NFS configuration file may be added to or overriden by additional command-line options. t Cm nfs.client.nextdowndelay When an NFS server is not responding, this option specifies how long to wait (in seconds) between notifications. The default is 30 seconds. t Cm nfs.client.nfsiod_thread_max The maximum number of NFS client asynchronous I/O (nfsiod) threads to use. The default is 16. t Cm nfs.client.statfs_rate_limit The maximum number of times per second that an NFS client will send a "statfs" RPC request to an NFS server to retrieve up-to-date file system information. Requests for this information that occur faster than this rate will receive cached values. The default value is 10 times per second. A value of zero means no limit. t Cm nfs.client.is_mobile This option specifies if an NFS client is on a mobile machine. On a mobile machine, hard mounted file systems can be automatically forcibly unmounted if the following is met: l -bullet -compact t The mount was automounted. t The server is not responding. t No files are open for writing. t No files are memory mapped. t There are no dirty pages associated with the mount. .El The default for this option is to let the system detect this. A non-zero value for this option treats this machine as a laptop with respect to NFS behavior and allows automatic forcible unmounting of unresponsive volumes described above. Conversely, a zero value treats the machine as a desktop for traditional NFS behavior, where hard mounts never time out. Setting a zero value is useful for laptops that are being used as desktops. t Cm nfs.lockd.grace_period This option specifies the grace period (in seconds) during which lockd will only accept requests from hosts which are reclaiming locks which existed before the restart. The default value is 45 seconds. t Cm nfs.lockd.host_monitor_cache_timeout This option tells rpc.lockd how long (in seconds) to cache state records for monitored hosts. Setting it to zero will disable the cache which will make lock and unlock requests from a single client more expensive because of additional interaction with the client's statd. The default value is 60 seconds. t Cm nfs.lockd.port This option can be used to specify a port that the NFS LOCK service (lockd) should be available on. The default value is unspecified, which means that any available port will be used. t Cm nfs.lockd.send_using_tcp This option tells lockd to use TCP sockets when contacting other hosts. The default value is 0, which means lockd will use UDP. t Cm nfs.lockd.send_using_mnt_transport This option tells lockd to use the socket type of the corresponding nfs mount. Locking a file on a udp mount will use udp and locking a file on a nfs mount mounted with tcp will use tcp. Note that nfs.lockd.send_using_tcp will override this option. The default value is 0, which means lockd will use UDP provided nfs.lockd.send_using_tcp is not set or set to 0. t Cm nfs.lockd.shutdown_delay_client This option species how long (in seconds) the lockd daemon should remain running after the unmounting of the last NFS file system for which it may need to perform file locking requests. The default value is 180 seconds. (Note: lockd may also remain running if it is needed by the NFS server.) t Cm nfs.lockd.shutdown_delay_server This option species how long (in seconds) the lockd daemon should remain running after the NFS server daemon, .Xr nfsd 8 , is stopped. The default value is 180 seconds. (Note: lockd may also remain running if it is needed by any NFS file system mounts.) t Cm nfs.lockd.tcp This option specifies whether the LOCK server should support connections using TCP. The default value is 1 (enabled). t Cm nfs.lockd.udp This option specifies whether the LOCK server should support connections using UDP. The default value is 1 (enabled). t Cm nfs.lockd.verbose This option controls how much logging lockd performs. It currently maps directly to the .Xr rpc.lockd 8 .Fl d Ar debug_level option. The default value is 0. t Cm nfs.server.async This option specifies that the NFS server should report unstable writes as stable writes. The default is 0 (off). While enabling this option can improve write performance, it will also put data integrity at risk because the NFS client will be told that data is on stable storage before it actually is. The data may be lost if the NFS server crashes. t Cm nfs.server.bonjour This option controls whether the NFS service is advertised via Bonjour. The default value is 1 (on). t Cm nfs.server.bonjour.local_domain_only This option controls whether the NFS service's Bonjour advertisement is restricted to the local domain only. The default value is 0 (off). t Cm nfs.server.export_hash_size This option specifies the size of the NFS server export hash table. The default value is 64. t Cm nfs.server.fsevents This option controls whether the NFS server will generate fsevents for operations performed on an exported file system. The default value is 1 (enabled). t Cm nfs.server.mount.port This option can be used to specify a port that the MOUNT service (mountd) should be available on. The default value is unspecified, which means that any available port will be used. t Cm nfs.server.mount.regular_files This option controls whether MOUNT requests for non-directory objects will be allowed. The default value is 0 (off). t Cm nfs.server.mount.require_resv_port This option controls whether MOUNT requests are required to originate from a reserved port (port < 1024). The default value is 1 (yes). Many NFS server implementations require this because of the false belief that this requirement increases security. t Cm nfs.server.nfsd_threads This option controls how many NFS server (nfsd) threads are made available to service NFS requests. The default value is 8. t Cm nfs.server.port This option can be used to specify a port that the NFS service (nfsd) should be available on. The default value is 2049. t Cm nfs.server.reqcache_size This option specifies the size of the NFS server request cache. The default value is 64. t Cm nfs.server.request_queue_length This option specifies the maximum number of NFS requests that the NFS server can queue up internally on the NFS server's UDP socket. The default value is 128. Note: using a large value risks getting all the mbufs in the system placed on that queue which can cause all networking to hang. t Cm nfs.server.require_resv_port This option controls whether NFS requests are required to originate from a reserved port (port < 1024). The default value is 0 (no). Many NFS server implementations require this because of the false belief that this requirement increases security. t Cm nfs.server.rquota.port This option can be used to specify a port that the RQUOTA service (rquotad) should be available on. The default value is unspecified, which means that any available port will be used. t Cm nfs.server.rquota.tcp This option specifies whether the RQUOTA server should support connections using TCP. The default value is 1 (enabled). t Cm nfs.server.rquota.udp This option specifies whether the RQUOTA server should support connections using UDP. The default value is 1 (enabled). t Cm nfs.server.tcp This option specifies whether the NFS server should support connections using TCP. The default value is 1 (enabled). t Cm nfs.server.udp This option specifies whether the NFS server should support connections using UDP. The default value is 1 (enabled). t Cm nfs.server.user_stats This option controls whether the NFS server maintains active user statistics. The default value is 1 (enabled). t Cm nfs.server.verbose This option controls how much logging nfsd performs. The default value is 0 - where only messages up to priority LOG_WARNING are logged. Setting the verbose level to 1 will add LOG_NOTICE messages which includes logging failed mount attempts. A verbose level of 2 will increase the log level to LOG_INFO which includes logging successful mount attempts. A log level of 3 or more will add LOG_DEBUG messages and cause increasing amounts of debug information to be logged. nfsd's verbose level can also be adjusted temporarily using the command: .Cm nfsd verbose . Note: the .Xr syslog 8 configuration may need to be adjusted in order to see the increased verbosity. t Cm nfs.server.wg_delay This option controls how long (in microseconds) NFSv2 writes will be gathered up before being processed. The default value is 1000. Setting this option to 0 will disable write gathering for NFSv2. t Cm nfs.server.wg_delay_v3 This option controls how long (in microseconds) NFSv3 writes will be gathered up before being processed. The default value is 0 (disabled). NFSv3's support of unstable writes effectively eliminates the need for doing write gathering to increase performance. t Cm nfs.statd.port This option can be used to specify a port that the STATUS service (statd) should be available on. The default value is unspecified, which means that any available port will be used. t Cm nfs.statd.send_using_tcp This option tells statd to use TCP sockets when contacting other hosts. The default value is 0, which means statd will use UDP. t Cm nfs.statd.simu_crash_allowed This option controls whether statd allows SM_SIMU_CRASH requests. The default value is 0 (not allowed). t Cm nfs.statd.tcp This option specifies whether the STATUS server should support connections using TCP. The default value is 1 (enabled). t Cm nfs.statd.udp This option specifies whether the STATUS server should support connections using UDP. The default value is 1 (enabled). t Cm nfs.statd.verbose This option controls how much logging statd performs. The default value is 0. .El .Sh FILES l -tag -width "/etc/nfs.conf" -compact t Pa /etc/nfs.conf The .Tn NFS configuration file. .El .Sh SEE ALSO .Xr nfsd 8 , .Xr rpc.lockd 8 , .Xr rpc.rquotad 8 , .Xr rpc.statd 8 , .Xr mount_nfs 8 .Sh BUGS \" Document known, unremedied bugs
.Sh HISTORY \" Document history if command behaves in a unique manner