|
 |
cf6d5cf |
commit ca43d9d8c7bdbd067fb1fb4fa9d6e055f4d34ce5
|
|
|
cf6d5cf |
Author: Steve Dickson <steved@redhat.com>
|
|
|
cf6d5cf |
Date: Fri Jan 4 09:27:35 2008 -0500
|
|
|
cf6d5cf |
|
|
|
cf6d5cf |
Incorporated Chuck Lever's and Don Domingo's changes to the
|
|
|
cf6d5cf |
nfs(5) manual page.
|
|
|
c0c131b |
|
|
|
cf6d5cf |
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
|
|
|
cf6d5cf |
Signed-off-by: Don Domingo <ddomingo@redhat.com>
|
|
|
cf6d5cf |
Signed-off-by: Steve Dickson <steved@redhat.com>
|
|
|
cf6d5cf |
|
|
|
c0c131b |
diff -up nfs-utils-1.1.1/utils/mount/nfs.man.save nfs-utils-1.1.1/utils/mount/nfs.man
|
|
|
c0c131b |
--- nfs-utils-1.1.1/utils/mount/nfs.man.save 2007-10-18 23:07:28.000000000 -0400
|
|
|
c0c131b |
+++ nfs-utils-1.1.1/utils/mount/nfs.man 2008-01-05 08:34:22.000000000 -0500
|
|
|
c0c131b |
@@ -1,510 +1,1233 @@
|
|
|
cf6d5cf |
-.\" nfs.5 "Rick Sladkey" <jrs@world.std.com>
|
|
|
cf6d5cf |
-.\" Wed Feb 8 12:52:42 1995, faith@cs.unc.edu: updates for Ross Biro's
|
|
|
cf6d5cf |
-.\" patches. "
|
|
|
cf6d5cf |
-.TH NFS 5 "20 November 1993" "Linux 0.99" "Linux Programmer's Manual"
|
|
|
cf6d5cf |
+.\"@(#)nfs.5"
|
|
|
cf6d5cf |
+.TH NFS 5 "2 November 2007"
|
|
|
cf6d5cf |
.SH NAME
|
|
|
cf6d5cf |
-nfs \- nfs and nfs4 fstab format and options
|
|
|
cf6d5cf |
+nfs \- fstab format and options for the
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file systems
|
|
|
cf6d5cf |
.SH SYNOPSIS
|
|
|
cf6d5cf |
-.B /etc/fstab
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
.SH DESCRIPTION
|
|
|
cf6d5cf |
-The
|
|
|
cf6d5cf |
-.I fstab
|
|
|
cf6d5cf |
-file contains information about which filesystems
|
|
|
cf6d5cf |
-to mount where and with what options.
|
|
|
cf6d5cf |
-For NFS mounts, it contains the server name and
|
|
|
cf6d5cf |
-exported server directory to mount from,
|
|
|
cf6d5cf |
-the local directory that is the mount point,
|
|
|
cf6d5cf |
-and the NFS specific options that control
|
|
|
cf6d5cf |
-the way the filesystem is mounted.
|
|
|
cf6d5cf |
+NFS is an Internet Standard protocol
|
|
|
cf6d5cf |
+created by Sun Microsystems in 1984. NFS was developed
|
|
|
cf6d5cf |
+to allow file sharing between systems residing
|
|
|
cf6d5cf |
+on a local area network.
|
|
|
cf6d5cf |
+The Linux NFS client supports three versions
|
|
|
cf6d5cf |
+of the NFS protocol:
|
|
|
cf6d5cf |
+NFS version 2 [RFC1094],
|
|
|
cf6d5cf |
+NFS version 3 [RFC1813],
|
|
|
cf6d5cf |
+and NFS version 4 [RFC3530].
|
|
|
cf6d5cf |
.P
|
|
|
cf6d5cf |
-Three different versions of the NFS protocol are
|
|
|
cf6d5cf |
-supported by the Linux NFS client:
|
|
|
cf6d5cf |
-NFS version 2, NFS version 3, and NFS version 4.
|
|
|
cf6d5cf |
-To mount via NFS version 2, use the
|
|
|
cf6d5cf |
-.BR nfs
|
|
|
cf6d5cf |
-file system type and specify
|
|
|
cf6d5cf |
-.BR nfsvers=2 .
|
|
|
cf6d5cf |
-To mount via NFS version 3, use the
|
|
|
cf6d5cf |
-.BR nfs
|
|
|
cf6d5cf |
-file system type and specify
|
|
|
cf6d5cf |
-.BR nfsvers=3 .
|
|
|
cf6d5cf |
-Version 3 is the default protocol version for the
|
|
|
cf6d5cf |
-.BR nfs
|
|
|
cf6d5cf |
-file system type when
|
|
|
cf6d5cf |
-.BR nfsvers=
|
|
|
cf6d5cf |
-is not specified on the mount command and both client and server
|
|
|
cf6d5cf |
-support it.
|
|
|
cf6d5cf |
-To mount via NFS version 4, use the
|
|
|
cf6d5cf |
-.BR nfs4
|
|
|
cf6d5cf |
-file system type.
|
|
|
cf6d5cf |
The
|
|
|
cf6d5cf |
-.BR nfsvers=
|
|
|
cf6d5cf |
-keyword is not supported for the
|
|
|
cf6d5cf |
-.BR nfs4
|
|
|
cf6d5cf |
-file system type.
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command attaches a file system to the system's
|
|
|
cf6d5cf |
+name space hierarchy at a given mount point.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
+file describes how
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+should assemble a system's file name hierarchy
|
|
|
cf6d5cf |
+from various independent file systems
|
|
|
cf6d5cf |
+(including file systems exported by NFS servers).
|
|
|
cf6d5cf |
+Each line in the
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
+file describes a single file system, its mount point,
|
|
|
cf6d5cf |
+and a set of default mount options for that mount point.
|
|
|
cf6d5cf |
.P
|
|
|
cf6d5cf |
-These file system types share similar mount options;
|
|
|
cf6d5cf |
-the differences are listed below.
|
|
|
cf6d5cf |
+For NFS file system mounts, a line in the
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
+file specifies the server name,
|
|
|
cf6d5cf |
+the path name of the exported server directory to mount,
|
|
|
cf6d5cf |
+the local directory that is the mount point,
|
|
|
cf6d5cf |
+the type of file system that is being mounted,
|
|
|
cf6d5cf |
+and a list of mount options that control
|
|
|
cf6d5cf |
+the way the filesystem is mounted and
|
|
|
cf6d5cf |
+how the NFS client behaves when accessing
|
|
|
cf6d5cf |
+files on this mount point.
|
|
|
cf6d5cf |
+The fifth and sixth fields on each line are not used
|
|
|
cf6d5cf |
+by NFS, thus conventionally each contain the digit zero. For example:
|
|
|
cf6d5cf |
.P
|
|
|
cf6d5cf |
-Here is an example from an \fI/etc/fstab\fP file for an NFSv3 mount
|
|
|
cf6d5cf |
-over TCP.
|
|
|
cf6d5cf |
-.sp
|
|
|
cf6d5cf |
-.nf
|
|
|
cf6d5cf |
-.ta 2.5i +0.75i +0.75i +1.0i
|
|
|
cf6d5cf |
-server:/usr/local/pub /pub nfs rsize=32768,wsize=32768,timeo=14,intr
|
|
|
cf6d5cf |
-.fi
|
|
|
cf6d5cf |
+.SP
|
|
|
cf6d5cf |
+.NF
|
|
|
cf6d5cf |
+.TA 2.5i +0.75i +0.75i +1.0i
|
|
|
cf6d5cf |
+ server:path /mountpoint fstype option,option,... 0 0
|
|
|
cf6d5cf |
+.FI
|
|
|
cf6d5cf |
.P
|
|
|
cf6d5cf |
-Here is an example for an NFSv4 mount over TCP using Kerberos
|
|
|
cf6d5cf |
-5 mutual authentication.
|
|
|
cf6d5cf |
-.sp
|
|
|
cf6d5cf |
-.nf
|
|
|
cf6d5cf |
-.ta 2.5i +0.75i +0.75i +1.0i
|
|
|
cf6d5cf |
-server:/usr/local/pub /pub nfs4 proto=tcp,sec=krb5,hard,intr
|
|
|
cf6d5cf |
-.fi
|
|
|
cf6d5cf |
+The server's hostname and export pathname
|
|
|
cf6d5cf |
+are separated by a colon, while
|
|
|
cf6d5cf |
+the mount options are separated by commas. The remaining fields
|
|
|
cf6d5cf |
+are separated by blanks or tabs.
|
|
|
cf6d5cf |
+The server's hostname can be an unqualified hostname,
|
|
|
cf6d5cf |
+a fully qualified domain name,
|
|
|
cf6d5cf |
+or a dotted quad IPv4 address.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.I fstype
|
|
|
cf6d5cf |
+field contains either "nfs" (for version 2 or version 3 NFS mounts)
|
|
|
cf6d5cf |
+or "nfs4" (for NFS version 4 mounts).
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file system types share similar mount options,
|
|
|
cf6d5cf |
+which are described below.
|
|
|
cf6d5cf |
+.SH "MOUNT OPTIONS"
|
|
|
cf6d5cf |
+Refer to
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+for a description of generic mount options
|
|
|
cf6d5cf |
+available for all file systems. If you do not need to
|
|
|
cf6d5cf |
+specify any mount options, use the generic option
|
|
|
cf6d5cf |
+.B defaults
|
|
|
cf6d5cf |
+in
|
|
|
cf6d5cf |
+.IR /etc/fstab .
|
|
|
cf6d5cf |
+.
|
|
|
cf6d5cf |
.DT
|
|
|
cf6d5cf |
-.SS Options for the nfs file system type
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I rsize=n
|
|
|
cf6d5cf |
-The number of bytes NFS uses when reading files from an NFS server.
|
|
|
cf6d5cf |
-The rsize is negotiated between the server and client to determine
|
|
|
cf6d5cf |
-the largest block size that both can support.
|
|
|
cf6d5cf |
-The value specified by this option is the maximum size that could
|
|
|
cf6d5cf |
-be used; however, the actual size used may be smaller.
|
|
|
cf6d5cf |
-Note: Setting this size to a value less than the largest supported
|
|
|
cf6d5cf |
-block size will adversely affect performance.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I wsize=n
|
|
|
cf6d5cf |
-The number of bytes NFS uses when writing files to an NFS server.
|
|
|
cf6d5cf |
-The wsize is negotiated between the server and client to determine
|
|
|
cf6d5cf |
-the largest block size that both can support.
|
|
|
cf6d5cf |
-The value specified by this option is the maximum size that could
|
|
|
cf6d5cf |
-be used; however, the actual size used may be smaller.
|
|
|
cf6d5cf |
-Note: Setting this size to a value less than the largest supported
|
|
|
cf6d5cf |
-block size will adversely affect performance.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I timeo=n
|
|
|
cf6d5cf |
-The value in tenths of a second before sending the
|
|
|
cf6d5cf |
-first retransmission after an RPC timeout.
|
|
|
cf6d5cf |
-The default value is 7 tenths of a second. After the first timeout,
|
|
|
cf6d5cf |
-the timeout is doubled after each successive timeout until a maximum
|
|
|
cf6d5cf |
-timeout of 60 seconds is reached or the enough retransmissions
|
|
|
cf6d5cf |
-have occured to cause a major timeout. Then, if the filesystem
|
|
|
cf6d5cf |
-is hard mounted, each new timeout cascade restarts at twice the
|
|
|
cf6d5cf |
-initial value of the previous cascade, again doubling at each
|
|
|
cf6d5cf |
-retransmission. The maximum timeout is always 60 seconds.
|
|
|
cf6d5cf |
-Better overall performance may be achieved by increasing the
|
|
|
cf6d5cf |
-timeout when mounting on a busy network, to a slow server, or through
|
|
|
cf6d5cf |
-several routers or gateways.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I retrans=n
|
|
|
cf6d5cf |
-The number of minor timeouts and retransmissions that must occur before
|
|
|
cf6d5cf |
-a major timeout occurs. The default is 3 timeouts. When a major timeout
|
|
|
cf6d5cf |
-occurs, the file operation is either aborted or a "server not responding"
|
|
|
cf6d5cf |
-message is printed on the console.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acregmin=n
|
|
|
cf6d5cf |
-The minimum time in seconds that attributes of a regular file should
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 3 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acregmax=n
|
|
|
cf6d5cf |
-The maximum time in seconds that attributes of a regular file can
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 60 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acdirmin=n
|
|
|
cf6d5cf |
-The minimum time in seconds that attributes of a directory should
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 30 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acdirmax=n
|
|
|
cf6d5cf |
-The maximum time in seconds that attributes of a directory can
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 60 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I actimeo=n
|
|
|
cf6d5cf |
-Using actimeo sets all of
|
|
|
cf6d5cf |
-.I acregmin,
|
|
|
cf6d5cf |
-.I acregmax,
|
|
|
cf6d5cf |
-.I acdirmin,
|
|
|
cf6d5cf |
+.SS "Valid options for either the nfs or nfs4 file system type"
|
|
|
cf6d5cf |
+These options are valid to use when mounting either
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+or
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file system types.
|
|
|
cf6d5cf |
+They imply the same behavior
|
|
|
cf6d5cf |
+and have the same default for both file system types.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR soft " / " hard
|
|
|
cf6d5cf |
+Determines the recovery behavior of the NFS client
|
|
|
cf6d5cf |
+after an NFS request times out.
|
|
|
cf6d5cf |
+If neither option is specified (or if the
|
|
|
cf6d5cf |
+.B hard
|
|
|
cf6d5cf |
+option is specified), NFS requests are retried indefinitely.
|
|
|
cf6d5cf |
+If the
|
|
|
cf6d5cf |
+.B soft
|
|
|
cf6d5cf |
+option is specified, then the NFS client fails an NFS request
|
|
|
cf6d5cf |
+after
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
+retransmissions have been sent,
|
|
|
cf6d5cf |
+causing the NFS client to return an error
|
|
|
cf6d5cf |
+to the calling application.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+.I NB:
|
|
|
cf6d5cf |
+A so-called "soft" timeout can cause
|
|
|
cf6d5cf |
+silent data corruption in certain cases. As such, use the
|
|
|
cf6d5cf |
+.B soft
|
|
|
cf6d5cf |
+option only when client responsiveness
|
|
|
cf6d5cf |
+is more important than data integrity.
|
|
|
cf6d5cf |
+Using NFS over TCP or increasing the value of the
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
+option may mitigate some of the risks of using the
|
|
|
cf6d5cf |
+.B soft
|
|
|
cf6d5cf |
+option.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI timeo= n
|
|
|
cf6d5cf |
+The time (in tenths of a second) the NFS client waits for a
|
|
|
cf6d5cf |
+response before it retries an NFS request. If this
|
|
|
cf6d5cf |
+option is not specified, requests are retried after
|
|
|
cf6d5cf |
+60 seconds for NFS over TCP, and are retried after 7/10 of a second for
|
|
|
cf6d5cf |
+NFS over UDP.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+For NFS over TCP, the client uses a fixed timeout, as specified by the
|
|
|
cf6d5cf |
+.B timeo
|
|
|
cf6d5cf |
+option. However, for NFS over UDP, the client uses an adaptive
|
|
|
cf6d5cf |
+algorithm to estimate an appropriate timeout value for frequently used
|
|
|
cf6d5cf |
+request types (such as READ and WRITE requests), but uses the
|
|
|
cf6d5cf |
+.B timeo
|
|
|
cf6d5cf |
+setting for infrequently used request types (such as FSINFO requests).
|
|
|
cf6d5cf |
+After each retransmission, the NFS client doubles the timeout for that
|
|
|
cf6d5cf |
+request, up to a maximum timeout length of 60 seconds.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI retrans= n
|
|
|
cf6d5cf |
+The number of times the NFS client retries a request before
|
|
|
cf6d5cf |
+it attempts further recovery action. If the
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
+option is not specified, the NFS client retries each request
|
|
|
cf6d5cf |
+three times.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+The NFS client generates a "server not responding" message
|
|
|
cf6d5cf |
+after
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
+retries, then attempts further recovery (depending on whether the
|
|
|
cf6d5cf |
+.B hard
|
|
|
cf6d5cf |
+mount option is in effect).
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI rsize= n
|
|
|
cf6d5cf |
+The maximum number of bytes in each network READ request
|
|
|
cf6d5cf |
+that the NFS client can receive when reading data from a file
|
|
|
cf6d5cf |
+on an NFS server.
|
|
|
cf6d5cf |
+The actual data payload size of each NFS READ request is equal to
|
|
|
cf6d5cf |
+or smaller than the
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+setting. The largest read payload supported by the Linux NFS client
|
|
|
cf6d5cf |
+is 1,048,576 bytes (one megabyte).
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+value is a positive integral multiple of 1024.
|
|
|
cf6d5cf |
+Specified
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+values lower than 1024 are replaced with 4096; values larger than
|
|
|
cf6d5cf |
+1048576 are replaced with 1048576. If a specified value is within the supported
|
|
|
cf6d5cf |
+range but not a multiple of 1024, it is rounded down to the nearest
|
|
|
cf6d5cf |
+multiple of 1024.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If an
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+value is not specified, or if the specified
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+value is larger than the maximum that either client or server can support,
|
|
|
cf6d5cf |
+the client and server negotiate the largest
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+value that they can both support.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+mount option as specified on the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command line appears in the
|
|
|
cf6d5cf |
+.I /etc/mtab
|
|
|
cf6d5cf |
+file. However, the effective
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+value negotiated by the client and server is reported in the
|
|
|
cf6d5cf |
+.I /proc/mounts
|
|
|
cf6d5cf |
+file.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI wsize= n
|
|
|
cf6d5cf |
+The maximum number of bytes per network WRITE request
|
|
|
cf6d5cf |
+that the NFS client can send when writing data to a file
|
|
|
cf6d5cf |
+on an NFS server. The actual data payload size of each
|
|
|
cf6d5cf |
+NFS WRITE request is equal to
|
|
|
cf6d5cf |
+or smaller than the
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+setting. The largest write payload supported by the Linux NFS client
|
|
|
cf6d5cf |
+is 1,048,576 bytes (one megabyte).
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Similar to
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+, the
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+value is a positive integral multiple of 1024.
|
|
|
cf6d5cf |
+Specified
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+values lower than 1024 are replaced with 4096; values larger than
|
|
|
cf6d5cf |
+1048576 are replaced with 1048576. If a specified value is within the supported
|
|
|
cf6d5cf |
+range but not a multiple of 1024, it is rounded down to the nearest
|
|
|
cf6d5cf |
+multiple of 1024.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If a
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+value is not specified, or if the specified
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+value is larger than the maximum that either client or server can support,
|
|
|
cf6d5cf |
+the client and server negotiate the largest
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+value that they can both support.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+mount option as specified on the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command line appears in the
|
|
|
cf6d5cf |
+.I /etc/mtab
|
|
|
cf6d5cf |
+file. However, the effective
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+value negotiated by the client and server is reported in the
|
|
|
cf6d5cf |
+.I /proc/mounts
|
|
|
cf6d5cf |
+file.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR ac " / " noac
|
|
|
cf6d5cf |
+Selects whether the client may cache file attributes. If neither
|
|
|
cf6d5cf |
+option is specified (or if
|
|
|
cf6d5cf |
+.B ac
|
|
|
cf6d5cf |
+is specified), the client caches file
|
|
|
cf6d5cf |
+attributes.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+To improve performance, NFS clients cache file
|
|
|
cf6d5cf |
+attributes. Every few seconds, an NFS client checks the server's version of each
|
|
|
cf6d5cf |
+file's attributes for updates. Changes that occur on the server in
|
|
|
cf6d5cf |
+those small intervals remain undetected until the client checks the
|
|
|
cf6d5cf |
+server again. The
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option prevents clients from caching file
|
|
|
cf6d5cf |
+attributes so that applications can more quickly detect file changes
|
|
|
cf6d5cf |
+on the server.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+In addition to preventing the client from caching file attributes,
|
|
|
cf6d5cf |
+the
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option forces application writes to become synchronous so
|
|
|
cf6d5cf |
+that local changes to a file become visible on the server
|
|
|
cf6d5cf |
+immediately. That way, other clients can quickly detect recent
|
|
|
cf6d5cf |
+writes when they check the file's attributes.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Using the
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option provides greater cache coherence among NFS clients
|
|
|
cf6d5cf |
+accessing the same files,
|
|
|
cf6d5cf |
+but it extracts a significant performance penalty.
|
|
|
cf6d5cf |
+As such, judicious use of file locking is encouraged instead.
|
|
|
cf6d5cf |
+The DATA AND METADATA COHERENCE section contains a detailed discussion
|
|
|
cf6d5cf |
+of these trade-offs.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI acregmin= n
|
|
|
cf6d5cf |
+The minimum time (in seconds) that the NFS client caches
|
|
|
cf6d5cf |
+attributes of a regular file before it requests
|
|
|
cf6d5cf |
+fresh attribute information from a server.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses
|
|
|
cf6d5cf |
+a 3-second minimum.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI acregmax= n
|
|
|
cf6d5cf |
+The maximum time (in seconds) that the NFS client caches
|
|
|
cf6d5cf |
+attributes of a regular file before it requests
|
|
|
cf6d5cf |
+fresh attribute information from a server.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses
|
|
|
cf6d5cf |
+a 60-second maximum.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI acdirmin= n
|
|
|
cf6d5cf |
+The minimum time (in seconds) that the NFS client caches
|
|
|
cf6d5cf |
+attributes of a directory before it requests
|
|
|
cf6d5cf |
+fresh attribute information from a server.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses
|
|
|
cf6d5cf |
+a 30-second minimum.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI acdirmax= n
|
|
|
cf6d5cf |
+The maximum time (in seconds) that the NFS client caches
|
|
|
cf6d5cf |
+attributes of a directory before it requests
|
|
|
cf6d5cf |
+fresh attribute information from a server.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses
|
|
|
cf6d5cf |
+a 60-second maximum.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI actimeo= n
|
|
|
cf6d5cf |
+Using
|
|
|
cf6d5cf |
+.B actimeo
|
|
|
cf6d5cf |
+sets all of
|
|
|
cf6d5cf |
+.BR acregmin ,
|
|
|
cf6d5cf |
+.BR acregmax ,
|
|
|
cf6d5cf |
+.BR acdirmin ,
|
|
|
cf6d5cf |
and
|
|
|
cf6d5cf |
-.I acdirmax
|
|
|
cf6d5cf |
+.B acdirmax
|
|
|
cf6d5cf |
to the same value.
|
|
|
cf6d5cf |
-There is no default value.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses
|
|
|
cf6d5cf |
+the defaults for each of these options listed above.
|
|
|
cf6d5cf |
.TP 1.5i
|
|
|
cf6d5cf |
-.I retry=n
|
|
|
cf6d5cf |
-The number of minutes to retry an NFS mount operation
|
|
|
cf6d5cf |
+.BR bg " / " fg
|
|
|
cf6d5cf |
+Determines how the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command behaves if an attempt to mount an export fails.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B fg
|
|
|
cf6d5cf |
+option causes
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+to exit with an error status if any part of the mount request
|
|
|
cf6d5cf |
+times out or fails outright.
|
|
|
cf6d5cf |
+This is called a "foreground" mount,
|
|
|
cf6d5cf |
+and is the default behavior if neither the
|
|
|
cf6d5cf |
+.B fg
|
|
|
cf6d5cf |
+nor
|
|
|
cf6d5cf |
+.B bg
|
|
|
cf6d5cf |
+mount option is specified.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If the
|
|
|
cf6d5cf |
+.B bg
|
|
|
cf6d5cf |
+option is specified, a timeout or failure causes the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command to fork a child which continues to attempt
|
|
|
cf6d5cf |
+to mount the export.
|
|
|
cf6d5cf |
+The parent immediately returns with a zero exit code.
|
|
|
cf6d5cf |
+This is known as a "background" mount.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If the local mount point directory is missing, the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command acts as if the mount request timed out.
|
|
|
cf6d5cf |
+This permits nested NFS mounts specified in
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
+to proceed in any order during system initialization,
|
|
|
cf6d5cf |
+even if some NFS servers are not yet available.
|
|
|
cf6d5cf |
+Alternatively these issues can be addressed
|
|
|
cf6d5cf |
+using an automounter (refer to
|
|
|
cf6d5cf |
+.BR automount (8)
|
|
|
cf6d5cf |
+for details).
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI retry= n
|
|
|
cf6d5cf |
+The number of minutes that the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command retries an NFS mount operation
|
|
|
cf6d5cf |
in the foreground or background before giving up.
|
|
|
cf6d5cf |
-The default value for forground mounts is 2 minutes.
|
|
|
cf6d5cf |
-The default value for background mounts is 10000 minutes,
|
|
|
cf6d5cf |
-which is roughly one week.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I namlen=n
|
|
|
cf6d5cf |
-When an NFS server does not support version two of the
|
|
|
cf6d5cf |
-RPC mount protocol, this option can be used to specify
|
|
|
cf6d5cf |
-the maximum length of a filename that is supported on
|
|
|
cf6d5cf |
-the remote filesystem. This is used to support the
|
|
|
cf6d5cf |
-POSIX pathconf functions. The default is 255 characters.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I port=n
|
|
|
cf6d5cf |
-The numeric value of the port to connect to the NFS server on.
|
|
|
cf6d5cf |
-If the port number is 0 (the default) then query the
|
|
|
cf6d5cf |
-remote host's portmapper for the port number to use.
|
|
|
cf6d5cf |
-If the remote host's NFS daemon is not registered with
|
|
|
cf6d5cf |
-its portmapper, the standard NFS port number 2049 is
|
|
|
cf6d5cf |
-used instead.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I mountport=n
|
|
|
cf6d5cf |
-The numeric value of the
|
|
|
cf6d5cf |
-.B mountd
|
|
|
cf6d5cf |
-port.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I mounthost=name
|
|
|
cf6d5cf |
-The name of the host running
|
|
|
cf6d5cf |
-.B mountd .
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I mountprog=n
|
|
|
cf6d5cf |
-Use an alternate RPC program number to contact the
|
|
|
cf6d5cf |
-mount daemon on the remote host. This option is useful
|
|
|
cf6d5cf |
-for hosts that can run multiple NFS servers.
|
|
|
cf6d5cf |
-The default value is 100005 which is the standard RPC
|
|
|
cf6d5cf |
-mount daemon program number.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I mountvers=n
|
|
|
cf6d5cf |
-Use an alternate RPC version number to contact the
|
|
|
cf6d5cf |
-mount daemon on the remote host. This option is useful
|
|
|
cf6d5cf |
-for hosts that can run multiple NFS servers.
|
|
|
cf6d5cf |
-The default value depends on which kernel you are using.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nfsprog=n
|
|
|
cf6d5cf |
-Use an alternate RPC program number to contact the
|
|
|
cf6d5cf |
-NFS daemon on the remote host. This option is useful
|
|
|
cf6d5cf |
-for hosts that can run multiple NFS servers.
|
|
|
cf6d5cf |
-The default value is 100003 which is the standard RPC
|
|
|
cf6d5cf |
-NFS daemon program number.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nfsvers=n
|
|
|
cf6d5cf |
-Use an alternate RPC version number to contact the
|
|
|
cf6d5cf |
-NFS daemon on the remote host. This option is useful
|
|
|
cf6d5cf |
-for hosts that can run multiple NFS servers.
|
|
|
cf6d5cf |
-The default value depends on which kernel you are using.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I vers=n
|
|
|
cf6d5cf |
-vers is an alternative to nfsvers and is compatible with
|
|
|
cf6d5cf |
-many other operating systems.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nolock
|
|
|
cf6d5cf |
-Disable NFS locking. Do not start lockd.
|
|
|
cf6d5cf |
-This is appropriate for mounting the root filesystem or
|
|
|
cf6d5cf |
-.B /usr
|
|
|
cf6d5cf |
-or
|
|
|
cf6d5cf |
-.BR /var .
|
|
|
cf6d5cf |
-These filesystems are typically either read-only or not shared, and in
|
|
|
cf6d5cf |
-those cases, remote locking is not needed.
|
|
|
cf6d5cf |
-This also needs to be used with some old NFS servers
|
|
|
cf6d5cf |
-that don't support locking.
|
|
|
cf6d5cf |
-.br
|
|
|
cf6d5cf |
-Note that applications can still get locks on files, but the locks
|
|
|
cf6d5cf |
-only provide exclusion locally. Other clients mounting the same
|
|
|
cf6d5cf |
-filesystem will not be able to detect the locks.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I bg
|
|
|
cf6d5cf |
-If the first NFS mount attempt times out, retry the mount
|
|
|
cf6d5cf |
-in the background.
|
|
|
cf6d5cf |
-After a mount operation is backgrounded, all subsequent mounts
|
|
|
cf6d5cf |
-on the same NFS server will be backgrounded immediately, without
|
|
|
cf6d5cf |
-first attempting the mount.
|
|
|
cf6d5cf |
-A missing mount point is treated as a timeout,
|
|
|
cf6d5cf |
-to allow for nested NFS mounts.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I fg
|
|
|
cf6d5cf |
-If the first NFS mount attempt times out, retry the mount
|
|
|
cf6d5cf |
-in the foreground.
|
|
|
cf6d5cf |
-This is the complement of the
|
|
|
cf6d5cf |
-.I bg
|
|
|
cf6d5cf |
-option, and also the default behavior.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I soft
|
|
|
cf6d5cf |
-If an NFS file operation has a major timeout then report an I/O error to
|
|
|
cf6d5cf |
-the calling program.
|
|
|
cf6d5cf |
-The default is to continue retrying NFS file operations indefinitely.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I hard
|
|
|
cf6d5cf |
-If an NFS file operation has a major timeout then report
|
|
|
cf6d5cf |
-"server not responding" on the console and continue retrying indefinitely.
|
|
|
cf6d5cf |
-This is the default.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I intr
|
|
|
cf6d5cf |
-If an NFS file operation has a major timeout and it is hard mounted,
|
|
|
cf6d5cf |
-then allow signals to interupt the file operation and cause it to
|
|
|
cf6d5cf |
-return EINTR to the calling program. The default is to not
|
|
|
cf6d5cf |
-allow file operations to be interrupted.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I posix
|
|
|
cf6d5cf |
-Mount the NFS filesystem using POSIX semantics. This allows
|
|
|
cf6d5cf |
-an NFS filesystem to properly support the POSIX pathconf
|
|
|
cf6d5cf |
-command by querying the mount server for the maximum length
|
|
|
cf6d5cf |
-of a filename. To do this, the remote host must support version
|
|
|
cf6d5cf |
-two of the RPC mount protocol. Many NFS servers support only
|
|
|
cf6d5cf |
-version one.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nocto
|
|
|
cf6d5cf |
-Suppress the retrieval of new attributes when creating a file.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I noac
|
|
|
cf6d5cf |
-Disable all forms of attribute caching entirely. This extracts a
|
|
|
cf6d5cf |
-significant performance penalty but it allows two different NFS clients
|
|
|
cf6d5cf |
-to get reasonable results when both clients are actively
|
|
|
cf6d5cf |
-writing to a common export on the server.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I noacl
|
|
|
cf6d5cf |
-Disables Access Control List (ACL) processing.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I sec=mode
|
|
|
cf6d5cf |
-Set the security flavor for this mount to "mode".
|
|
|
cf6d5cf |
-The default setting is \f3sec=sys\f1, which uses local
|
|
|
cf6d5cf |
-unix uids and gids to authenticate NFS operations (AUTH_SYS).
|
|
|
cf6d5cf |
-Other currently supported settings are:
|
|
|
cf6d5cf |
-\f3sec=krb5\f1, which uses Kerberos V5 instead of local unix uids
|
|
|
cf6d5cf |
-and gids to authenticate users;
|
|
|
cf6d5cf |
-\f3sec=krb5i\f1, which uses Kerberos V5 for user authentication
|
|
|
cf6d5cf |
-and performs integrity checking of NFS operations using secure
|
|
|
cf6d5cf |
-checksums to prevent data tampering; and
|
|
|
cf6d5cf |
-\f3sec=krb5p\f1, which uses Kerberos V5 for user authentication
|
|
|
cf6d5cf |
-and integrity checking, and encrypts NFS traffic to prevent
|
|
|
cf6d5cf |
-traffic sniffing (this is the most secure setting).
|
|
|
cf6d5cf |
-Note that there is a performance penalty when using integrity
|
|
|
cf6d5cf |
-or privacy.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I tcp
|
|
|
cf6d5cf |
-Mount the NFS filesystem using the TCP protocol. This is the default
|
|
|
cf6d5cf |
-if it is supported by both client and server. Many NFS servers only
|
|
|
cf6d5cf |
-support UDP.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I udp
|
|
|
cf6d5cf |
-Mount the NFS filesystem using the UDP protocol.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nordirplus
|
|
|
cf6d5cf |
-Disables NFSv3 READDIRPLUS RPCs. Use this option when
|
|
|
cf6d5cf |
-mounting servers that don't support or have broken
|
|
|
cf6d5cf |
-READDIRPLUS implementations.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I nosharecache
|
|
|
cf6d5cf |
-As of kernel 2.6.18, it is no longer possible to mount the same
|
|
|
cf6d5cf |
-same filesystem with different mount options to a new mountpoint.
|
|
|
cf6d5cf |
-It was deemed unsafe to do so, since cached data cannot be shared
|
|
|
cf6d5cf |
-between the two mountpoints. In consequence, files or directories
|
|
|
cf6d5cf |
-that were common to both mountpoint subtrees could often be seen to
|
|
|
cf6d5cf |
-be out of sync following an update.
|
|
|
cf6d5cf |
-.br
|
|
|
cf6d5cf |
-This option allows administrators to select the pre-2.6.18 behaviour,
|
|
|
cf6d5cf |
-permitting the same filesystem to be mounted with different mount
|
|
|
cf6d5cf |
-options.
|
|
|
cf6d5cf |
-.br
|
|
|
cf6d5cf |
-.B Beware:
|
|
|
cf6d5cf |
-Use of this option is not recommended unless you are certain that there
|
|
|
cf6d5cf |
-are no hard links or subtrees of this mountpoint that are mounted
|
|
|
cf6d5cf |
-elsewhere.
|
|
|
cf6d5cf |
-.P
|
|
|
cf6d5cf |
-All of the non-value options have corresponding nooption forms.
|
|
|
cf6d5cf |
-For example, nointr means don't allow file operations to be
|
|
|
cf6d5cf |
-interrupted.
|
|
|
cf6d5cf |
-.SS Options for the nfs4 file system type
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I rsize=n
|
|
|
cf6d5cf |
-The number of bytes nfs4 uses when reading files from the server.
|
|
|
cf6d5cf |
-The rsize is negotiated between the server and client to determine
|
|
|
cf6d5cf |
-the largest block size that both can support.
|
|
|
cf6d5cf |
-The value specified by this option is the maximum size that could
|
|
|
cf6d5cf |
-be used; however, the actual size used may be smaller.
|
|
|
cf6d5cf |
-Note: Setting this size to a value less than the largest supported
|
|
|
cf6d5cf |
-block size will adversely affect performance.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I wsize=n
|
|
|
cf6d5cf |
-The number of bytes nfs4 uses when writing files to the server.
|
|
|
cf6d5cf |
-The wsize is negotiated between the server and client to determine
|
|
|
cf6d5cf |
-the largest block size that both can support.
|
|
|
cf6d5cf |
-The value specified by this option is the maximum size that could
|
|
|
cf6d5cf |
-be used; however, the actual size used may be smaller.
|
|
|
cf6d5cf |
-Note: Setting this size to a value less than the largest supported
|
|
|
cf6d5cf |
-block size will adversely affect performance.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I timeo=n
|
|
|
cf6d5cf |
-The value in tenths of a second before sending the
|
|
|
cf6d5cf |
-first retransmission after an RPC timeout.
|
|
|
cf6d5cf |
-The default value depends on whether
|
|
|
cf6d5cf |
-.IR proto=udp
|
|
|
cf6d5cf |
-or
|
|
|
cf6d5cf |
-.IR proto=tcp
|
|
|
cf6d5cf |
-is in effect (see below).
|
|
|
cf6d5cf |
-The default value for UDP is 7 tenths of a second.
|
|
|
cf6d5cf |
-The default value for TCP is 60 seconds.
|
|
|
cf6d5cf |
-After the first timeout,
|
|
|
cf6d5cf |
-the timeout is doubled after each successive timeout until a maximum
|
|
|
cf6d5cf |
-timeout of 60 seconds is reached or the enough retransmissions
|
|
|
cf6d5cf |
-have occured to cause a major timeout. Then, if the filesystem
|
|
|
cf6d5cf |
-is hard mounted, each new timeout cascade restarts at twice the
|
|
|
cf6d5cf |
-initial value of the previous cascade, again doubling at each
|
|
|
cf6d5cf |
-retransmission. The maximum timeout is always 60 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I retrans=n
|
|
|
cf6d5cf |
-The number of minor timeouts and retransmissions that must occur before
|
|
|
cf6d5cf |
-a major timeout occurs. The default is 5 timeouts for
|
|
|
cf6d5cf |
-.IR proto=udp
|
|
|
cf6d5cf |
-and 2 timeouts for
|
|
|
cf6d5cf |
-.IR proto=tcp .
|
|
|
cf6d5cf |
-When a major timeout
|
|
|
cf6d5cf |
-occurs, the file operation is either aborted or a "server not responding"
|
|
|
cf6d5cf |
-message is printed on the console.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acregmin=n
|
|
|
cf6d5cf |
-The minimum time in seconds that attributes of a regular file should
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 3 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acregmax=n
|
|
|
cf6d5cf |
-The maximum time in seconds that attributes of a regular file can
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 60 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acdirmin=n
|
|
|
cf6d5cf |
-The minimum time in seconds that attributes of a directory should
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 30 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I acdirmax=n
|
|
|
cf6d5cf |
-The maximum time in seconds that attributes of a directory can
|
|
|
cf6d5cf |
-be cached before requesting fresh information from a server.
|
|
|
cf6d5cf |
-The default is 60 seconds.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I actimeo=n
|
|
|
cf6d5cf |
-Using actimeo sets all of
|
|
|
cf6d5cf |
-.I acregmin,
|
|
|
cf6d5cf |
-.I acregmax,
|
|
|
cf6d5cf |
-.I acdirmin,
|
|
|
cf6d5cf |
+If this option is not specified, the default value for foreground mounts
|
|
|
cf6d5cf |
+is 2 minutes, and the default value for background mounts is 10000 minutes (80 minutes shy of one week).
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI sec= mode
|
|
|
cf6d5cf |
+The RPCGSS security flavor to use for accessing files on this mount point.
|
|
|
cf6d5cf |
+If the
|
|
|
cf6d5cf |
+.B sec
|
|
|
cf6d5cf |
+option is not specified, or if
|
|
|
cf6d5cf |
+.B sec=sys
|
|
|
cf6d5cf |
+is specified, the NFS client uses the AUTH_SYS security flavor
|
|
|
cf6d5cf |
+for all NFS requests on this mount point.
|
|
|
cf6d5cf |
+Valid security flavors are
|
|
|
cf6d5cf |
+.BR none ,
|
|
|
cf6d5cf |
+.BR sys ,
|
|
|
cf6d5cf |
+.BR krb5 ,
|
|
|
cf6d5cf |
+.BR krb5i ,
|
|
|
cf6d5cf |
+.BR krb5p ,
|
|
|
cf6d5cf |
+.BR lkey ,
|
|
|
cf6d5cf |
+.BR lkeyi ,
|
|
|
cf6d5cf |
+.BR lkeyp ,
|
|
|
cf6d5cf |
+.BR spkm ,
|
|
|
cf6d5cf |
+.BR spkmi ,
|
|
|
cf6d5cf |
and
|
|
|
cf6d5cf |
-.I acdirmax
|
|
|
cf6d5cf |
-to the same value.
|
|
|
cf6d5cf |
-There is no default value.
|
|
|
cf6d5cf |
+.BR spkmp .
|
|
|
cf6d5cf |
+Refer to the SECURITY CONSIDERATIONS section for details.
|
|
|
cf6d5cf |
.TP 1.5i
|
|
|
cf6d5cf |
-.I retry=n
|
|
|
cf6d5cf |
-The number of minutes to retry an NFS mount operation
|
|
|
cf6d5cf |
-in the foreground or background before giving up.
|
|
|
cf6d5cf |
-The default value for forground mounts is 2 minutes.
|
|
|
cf6d5cf |
-The default value for background mounts is 10000 minutes,
|
|
|
cf6d5cf |
-which is roughly one week.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I port=n
|
|
|
cf6d5cf |
-The numeric value of the port to connect to the NFS server on.
|
|
|
cf6d5cf |
-If the port number is 0 (the default) then query the
|
|
|
cf6d5cf |
-remote host's portmapper for the port number to use.
|
|
|
cf6d5cf |
-If the remote host's NFS daemon is not registered with
|
|
|
cf6d5cf |
-its portmapper, the standard NFS port number 2049 is
|
|
|
cf6d5cf |
-used instead.
|
|
|
cf6d5cf |
-.TP 1.5i
|
|
|
cf6d5cf |
-.I proto=n
|
|
|
cf6d5cf |
-Mount the NFS filesystem using a specific network protocol
|
|
|
cf6d5cf |
-instead of the default UDP protocol.
|
|
|
cf6d5cf |
-Many NFS version 4 servers only support TCP.
|
|
|
cf6d5cf |
-Valid protocol types are
|
|
|
cf6d5cf |
-.IR udp
|
|
|
cf6d5cf |
+.BR sharecache " / " nosharecache
|
|
|
cf6d5cf |
+Determines how the client's data cache and attribute cache are shared
|
|
|
cf6d5cf |
+when mounting the same export more than once concurrently. Using the
|
|
|
cf6d5cf |
+same cache reduces memory requirements on the client and presents
|
|
|
cf6d5cf |
+identical file contents to applications when the same remote file is
|
|
|
cf6d5cf |
+accessed via different mount points.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If neither option is specified, or if the
|
|
|
cf6d5cf |
+.B sharecache
|
|
|
cf6d5cf |
+option is
|
|
|
cf6d5cf |
+specified, then a single cache is used for all mount points that
|
|
|
cf6d5cf |
+access the same export. If the
|
|
|
cf6d5cf |
+.B nosharecache
|
|
|
cf6d5cf |
+option is specified,
|
|
|
cf6d5cf |
+then that mount point gets a unique cache. Note that when data and
|
|
|
cf6d5cf |
+attribute caches are shared, the mount options from the first mount
|
|
|
cf6d5cf |
+point take effect for subsequent concurrent mounts of the same export.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+As of kernel 2.6.18, the behavior specified by
|
|
|
cf6d5cf |
+.B nosharecache
|
|
|
cf6d5cf |
+is legacy caching behavior. This
|
|
|
cf6d5cf |
+is considered a data risk since multiple cached copies
|
|
|
cf6d5cf |
+of the same file on the same client can become out of sync
|
|
|
cf6d5cf |
+following a local update of one of the copies.
|
|
|
cf6d5cf |
+.SS "Valid options for the nfs file system type"
|
|
|
cf6d5cf |
+Use these options, along with the options in the above subsection,
|
|
|
cf6d5cf |
+for mounting the
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+file system type.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI proto= netid
|
|
|
cf6d5cf |
+The transport protocol used by the NFS client
|
|
|
cf6d5cf |
+to transmit requests to the NFS server for this mount point.
|
|
|
cf6d5cf |
+.I netid
|
|
|
cf6d5cf |
+can be either
|
|
|
cf6d5cf |
+.B udp
|
|
|
cf6d5cf |
+or
|
|
|
cf6d5cf |
+.BR tcp .
|
|
|
cf6d5cf |
+Each transport protocol uses different default
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
and
|
|
|
cf6d5cf |
-.IR tcp .
|
|
|
cf6d5cf |
+.B timeo
|
|
|
cf6d5cf |
+settings; refer to the description of these two mount options for details.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+In addition to controlling how the NFS client transmits requests to
|
|
|
cf6d5cf |
+the server, this mount option also controls how the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command communicates with the server's rpcbind and mountd services.
|
|
|
cf6d5cf |
+Specifying
|
|
|
cf6d5cf |
+.B proto=tcp
|
|
|
cf6d5cf |
+forces all traffic from the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command and the NFS client to use TCP.
|
|
|
cf6d5cf |
+Specifying
|
|
|
cf6d5cf |
+.B proto=udp
|
|
|
cf6d5cf |
+forces all traffic types to use UDP.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If the
|
|
|
cf6d5cf |
+.B proto
|
|
|
cf6d5cf |
+mount option is not specified, the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command discovers which protocols the server supports
|
|
|
cf6d5cf |
+and chooses an appropriate transport for each service.
|
|
|
cf6d5cf |
+Refer to the TRANSPORT METHODS section for more details.
|
|
|
cf6d5cf |
.TP 1.5i
|
|
|
c0c131b |
-.I clientaddr=n.n.n.n
|
|
|
c0c131b |
-Specifies a single IPv4 address in dotted-quad form that
|
|
|
c0c131b |
-the NFS client advertises to allow servers to perform
|
|
|
c0c131b |
-NFSv4 callback requests against files on this mount point.
|
|
|
c0c131b |
-If the server is not able to establish callback connections
|
|
|
c0c131b |
-to clients, performance may degrade, or accesses to
|
|
|
c0c131b |
-files may temporarily hang.
|
|
|
cf6d5cf |
+.B udp
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B udp
|
|
|
cf6d5cf |
+option is an alternative to specifying
|
|
|
cf6d5cf |
+.BR proto=udp.
|
|
|
cf6d5cf |
+It is included for compatibility with other operating systems.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.B tcp
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B tcp
|
|
|
cf6d5cf |
+option is an alternative to specifying
|
|
|
cf6d5cf |
+.BR proto=tcp.
|
|
|
cf6d5cf |
+It is included for compatibility with other operating systems.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI port= n
|
|
|
cf6d5cf |
+The numeric value of the server's NFS service port.
|
|
|
cf6d5cf |
+If the server's NFS service is not available on the specified port,
|
|
|
cf6d5cf |
+the mount request fails.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If this option is not specified, or if the specified port value is 0,
|
|
|
cf6d5cf |
+then the NFS client uses the NFS service port number
|
|
|
cf6d5cf |
+advertised by the server's rpcbind service.
|
|
|
cf6d5cf |
+The mount request fails if the server's rpcbind service is not available,
|
|
|
cf6d5cf |
+the server's NFS service is not registered with its rpcbind service,
|
|
|
cf6d5cf |
+or the server's NFS service is not available on the advertised port.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI mountport= n
|
|
|
cf6d5cf |
+The numeric value of the server's mountd port.
|
|
|
cf6d5cf |
+If the server's mountd service is not available on the specified port,
|
|
|
cf6d5cf |
+the mount request fails.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If this option is not specified,
|
|
|
cf6d5cf |
+or if the specified port value is 0, then the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command uses the mountd service port number
|
|
|
cf6d5cf |
+advertised by the server's rpcbind service.
|
|
|
cf6d5cf |
+The mount request fails if the server's rpcbind service is not available,
|
|
|
cf6d5cf |
+the server's mountd service is not registered with its rpcbind service,
|
|
|
cf6d5cf |
+or the server's mountd service is not available on the advertised port.
|
|
|
c0c131b |
.IP
|
|
|
cf6d5cf |
+This option can be used when mounting an NFS server
|
|
|
cf6d5cf |
+through a firewall that blocks the rpcbind protocol.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI mounthost= name
|
|
|
cf6d5cf |
+The hostname of the host running mountd.
|
|
|
c0c131b |
If this option is not specified, the
|
|
|
c0c131b |
.BR mount (8)
|
|
|
c0c131b |
-command attempts to discover an appropriate callback
|
|
|
c0c131b |
-address automatically.
|
|
|
cf6d5cf |
+command assumes that the mountd service runs
|
|
|
cf6d5cf |
+on the same host as the NFS service.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI mountvers= n
|
|
|
cf6d5cf |
+The RPC version number used to contact the server's mountd.
|
|
|
cf6d5cf |
+If this option is not specified, the client uses a version number
|
|
|
cf6d5cf |
+appropriate to the requested NFS version.
|
|
|
cf6d5cf |
+This option is useful when multiple NFS services
|
|
|
cf6d5cf |
+are running on the same remote server host.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI namlen= n
|
|
|
cf6d5cf |
+The maximum length of a pathname component on this mount.
|
|
|
cf6d5cf |
+If this option is not specified, the maximum length is negotiated
|
|
|
cf6d5cf |
+with the server. In most cases, this maximum length is 255 characters.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Some early versions of NFS did not support this negotiation.
|
|
|
cf6d5cf |
+Using this option ensures that
|
|
|
cf6d5cf |
+.BR pathconf (3)
|
|
|
cf6d5cf |
+reports the proper maximum component length to applications
|
|
|
cf6d5cf |
+in such cases.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI nfsvers= n
|
|
|
cf6d5cf |
+The NFS protocol version number used to contact the server's NFS service.
|
|
|
cf6d5cf |
+The Linux client supports version 2 and version 3 of the NFS protocol
|
|
|
cf6d5cf |
+when using the file system type
|
|
|
cf6d5cf |
+.BR nfs .
|
|
|
cf6d5cf |
+If the server does not support the requested version,
|
|
|
cf6d5cf |
+the mount request fails.
|
|
|
cf6d5cf |
+If this option is not specified, the client attempts to use version 3,
|
|
|
cf6d5cf |
+but negotiates the NFS version with the server if version 3 support
|
|
|
cf6d5cf |
+is not available.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI vers= n
|
|
|
cf6d5cf |
+This option is an alternative to the
|
|
|
cf6d5cf |
+.B nfsvers
|
|
|
cf6d5cf |
+option.
|
|
|
cf6d5cf |
+It is included for compatibility with other operating systems.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR lock " / " nolock
|
|
|
cf6d5cf |
+Selects whether to use the NLM sideband protocol to lock files on the server.
|
|
|
cf6d5cf |
+If neither option is specified (or if
|
|
|
cf6d5cf |
+.B lock
|
|
|
cf6d5cf |
+is specified), NLM locking is used for this mount point.
|
|
|
cf6d5cf |
+When using the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+option, applications can lock files,
|
|
|
cf6d5cf |
+but such locks provide exclusion only against other applications
|
|
|
cf6d5cf |
+running on the same client.
|
|
|
cf6d5cf |
+Remote applications are not affected by these locks.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+NLM locking must be disabled with the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+option when using NFS to mount
|
|
|
cf6d5cf |
+.I /var
|
|
|
cf6d5cf |
+because
|
|
|
cf6d5cf |
+.I /var
|
|
|
cf6d5cf |
+contains files used by the NLM implementation on Linux.
|
|
|
cf6d5cf |
+Using the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+option is also required when mounting exports on NFS servers
|
|
|
cf6d5cf |
+that do not support the NLM protocol.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR intr " / " nointr
|
|
|
cf6d5cf |
+Selects whether to allow signals to interrupt file operations
|
|
|
cf6d5cf |
+on this mount point. If neither option
|
|
|
cf6d5cf |
+is specified (or if
|
|
|
cf6d5cf |
+.B nointr
|
|
|
cf6d5cf |
+is specified),
|
|
|
cf6d5cf |
+signals do not interrupt NFS file operations. If
|
|
|
cf6d5cf |
+.B intr
|
|
|
cf6d5cf |
+is specified, system calls return EINTR if an in-progress NFS operation is interrupted by
|
|
|
cf6d5cf |
+a signal.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Using the
|
|
|
cf6d5cf |
+.B intr
|
|
|
cf6d5cf |
+option is preferred to using the
|
|
|
cf6d5cf |
+.B soft
|
|
|
cf6d5cf |
+option because it is significantly less likely to result in data corruption.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR cto " / " nocto
|
|
|
cf6d5cf |
+Selects whether to use close-to-open cache coherence semantics.
|
|
|
cf6d5cf |
+If neither option is specified (or if
|
|
|
cf6d5cf |
+.B cto
|
|
|
cf6d5cf |
+is specified), the client uses close-to-open
|
|
|
cf6d5cf |
+cache coherence semantics. If the
|
|
|
cf6d5cf |
+.B nocto
|
|
|
cf6d5cf |
+option is specified, the client uses a non-standard heuristic to determine when
|
|
|
cf6d5cf |
+files on the server have changed.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Using the
|
|
|
cf6d5cf |
+.B nocto
|
|
|
cf6d5cf |
+option may improve performance for read-only mounts,
|
|
|
cf6d5cf |
+but should be used only if the data on the server changes only occasionally.
|
|
|
cf6d5cf |
+The DATA AND METADATA COHERENCE section discusses the behavior
|
|
|
cf6d5cf |
+of this option in more detail.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR acl " / " noacl
|
|
|
cf6d5cf |
+Selects whether to use the NFSACL sideband protocol on this mount point.
|
|
|
cf6d5cf |
+The NFSACL sideband protocol is a proprietary protocol
|
|
|
cf6d5cf |
+implemented in Solaris that manages Access Control Lists. NFSACL was never
|
|
|
cf6d5cf |
+made a standard part of the NFS protocol specification.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If neither
|
|
|
cf6d5cf |
+.B acl
|
|
|
cf6d5cf |
+nor
|
|
|
cf6d5cf |
+.B noacl
|
|
|
cf6d5cf |
+option is specified,
|
|
|
cf6d5cf |
+the NFS client negotiates with the server
|
|
|
cf6d5cf |
+to see if the NFSACL protocol is supported,
|
|
|
cf6d5cf |
+and uses it if the server supports it.
|
|
|
cf6d5cf |
+Disabling the NFSACL sideband protocol may be necessary
|
|
|
cf6d5cf |
+if the negotiation causes problems on the client or server.
|
|
|
cf6d5cf |
+Refer to the SECURITY CONSIDERATIONS section for more details.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR rdirplus " / " nordirplus
|
|
|
cf6d5cf |
+Selects whether to use NFS version 3 READDIRPLUS requests.
|
|
|
cf6d5cf |
+If this option is not specified, the NFS client uses READDIRPLUS requests
|
|
|
cf6d5cf |
+on NFS version 3 mounts to read small directories.
|
|
|
cf6d5cf |
+Some applications perform better if the client uses only READDIR requests
|
|
|
cf6d5cf |
+for all directories.
|
|
|
cf6d5cf |
+.SS "Valid options for the nfs4 file system type"
|
|
|
cf6d5cf |
+Use these options, along with the options in the first subsection above,
|
|
|
cf6d5cf |
+for mounting the
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file system type.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI proto= netid
|
|
|
cf6d5cf |
+The transport protocol used by the NFS client
|
|
|
cf6d5cf |
+to transmit requests to the NFS server for this mount point.
|
|
|
cf6d5cf |
+.I netid
|
|
|
cf6d5cf |
+can be either
|
|
|
cf6d5cf |
+.B udp
|
|
|
cf6d5cf |
+or
|
|
|
cf6d5cf |
+.BR tcp .
|
|
|
cf6d5cf |
+All NFS version 4 servers are required to support TCP,
|
|
|
cf6d5cf |
+so if this mount option is not specified, the NFS version 4 client
|
|
|
cf6d5cf |
+uses the TCP transport protocol.
|
|
|
cf6d5cf |
+Refer to the TRANSPORT METHODS section for more details.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI port= n
|
|
|
cf6d5cf |
+The numeric value of the server's NFS service port.
|
|
|
cf6d5cf |
+If the server's NFS service is not available on the specified port,
|
|
|
cf6d5cf |
+the mount request fails.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If this mount option is not specified,
|
|
|
cf6d5cf |
+the NFS client uses the standard NFS port number of 2049
|
|
|
cf6d5cf |
+without first checking the server's rpcbind service.
|
|
|
cf6d5cf |
+This allows an NFS version 4 client to contact an NFS version 4
|
|
|
cf6d5cf |
+server through a firewall that may block rpcbind requests.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If the specified port value is 0,
|
|
|
cf6d5cf |
+then the NFS client uses the NFS service port number
|
|
|
cf6d5cf |
+advertised by the server's rpcbind service.
|
|
|
cf6d5cf |
+The mount request fails if the server's rpcbind service is not available,
|
|
|
cf6d5cf |
+the server's NFS service is not registered with its rpcbind service,
|
|
|
cf6d5cf |
+or the server's NFS service is not available on the advertised port.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR intr " / " nointr
|
|
|
cf6d5cf |
+Selects whether to allow signals to interrupt file operations
|
|
|
cf6d5cf |
+on this mount point. If neither option is specified (or if
|
|
|
cf6d5cf |
+.B intr
|
|
|
cf6d5cf |
+is specified), system calls return EINTR if an in-progress NFS operation
|
|
|
cf6d5cf |
+is interrupted by a signal. If
|
|
|
cf6d5cf |
+.B nointr
|
|
|
cf6d5cf |
+is specified, signals do not
|
|
|
cf6d5cf |
+interrupt NFS operations.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Using the
|
|
|
cf6d5cf |
+.B intr
|
|
|
cf6d5cf |
+option is preferred to using the
|
|
|
cf6d5cf |
+.B soft
|
|
|
cf6d5cf |
+option because it is significantly less likely to result in data corruption.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BR cto " / " nocto
|
|
|
cf6d5cf |
+Selects whether to use close-to-open cache coherence semantics
|
|
|
cf6d5cf |
+for NFS directories on this mount point.
|
|
|
cf6d5cf |
+If neither
|
|
|
cf6d5cf |
+.B cto
|
|
|
cf6d5cf |
+nor
|
|
|
cf6d5cf |
+.B nocto
|
|
|
cf6d5cf |
+is specified,
|
|
|
cf6d5cf |
+the default is to use close-to-open cache coherence
|
|
|
cf6d5cf |
+semantics for directories.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+File data caching behavior is not affected by this option.
|
|
|
cf6d5cf |
+The DATA AND METADATA COHERENCE section discusses
|
|
|
cf6d5cf |
+the behavior of this option in more detail.
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
+.BI clientaddr= n.n.n.n
|
|
|
cf6d5cf |
+Specifies a single IPv4 address (in dotted-quad form)
|
|
|
cf6d5cf |
+that the NFS client advertises to allow servers
|
|
|
cf6d5cf |
+to perform NFS version 4 callback requests against
|
|
|
cf6d5cf |
+files on this mount point. If the server is unable to
|
|
|
cf6d5cf |
+establish callback connections to clients, performance
|
|
|
cf6d5cf |
+may degrade, or accesses to files may temporarily hang.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+If this option is not specified, the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command attempts to discover an appropriate callback address automatically.
|
|
|
c0c131b |
The automatic discovery process is not perfect, however.
|
|
|
c0c131b |
-In the presence of multiple client network interfaces, special
|
|
|
c0c131b |
-routing policies, or atypical network topologies, the exact
|
|
|
c0c131b |
-address to use for callbacks may be nontrivial to determine,
|
|
|
c0c131b |
-and should be explicitly set using this mount option.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I sec=mode
|
|
|
c0c131b |
-Same as \f3sec=mode\f1 for the nfs filesystem type (see above).
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I bg
|
|
|
c0c131b |
-If an NFS mount attempt times out, retry the mount
|
|
|
c0c131b |
-in the background.
|
|
|
c0c131b |
-After a mount operation is backgrounded, all subsequent mounts
|
|
|
c0c131b |
-on the same NFS server will be backgrounded immediately, without
|
|
|
c0c131b |
-first attempting the mount.
|
|
|
c0c131b |
-A missing mount point is treated as a timeout,
|
|
|
c0c131b |
-to allow for nested NFS mounts.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I fg
|
|
|
c0c131b |
-If the first NFS mount attempt times out, retry the mount
|
|
|
c0c131b |
-in the foreground.
|
|
|
c0c131b |
-This is the complement of the
|
|
|
c0c131b |
-.I bg
|
|
|
c0c131b |
-option, and also the default behavior.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I soft
|
|
|
c0c131b |
-If an NFS file operation has a major timeout then report an I/O error to
|
|
|
c0c131b |
-the calling program.
|
|
|
c0c131b |
-The default is to continue retrying NFS file operations indefinitely.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I hard
|
|
|
c0c131b |
-If an NFS file operation has a major timeout then report
|
|
|
c0c131b |
-"server not responding" on the console and continue retrying indefinitely.
|
|
|
c0c131b |
-This is the default.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I intr
|
|
|
c0c131b |
-If an NFS file operation has a major timeout and it is hard mounted,
|
|
|
c0c131b |
-then allow signals to interupt the file operation and cause it to
|
|
|
c0c131b |
-return EINTR to the calling program. The default is to not
|
|
|
c0c131b |
-allow file operations to be interrupted.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I nocto
|
|
|
c0c131b |
-Suppress the retrieval of new attributes when creating a file.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I noac
|
|
|
c0c131b |
-Disable attribute caching, and force synchronous writes.
|
|
|
c0c131b |
-This extracts a
|
|
|
c0c131b |
-server performance penalty but it allows two different NFS clients
|
|
|
c0c131b |
-to get reasonable good results when both clients are actively
|
|
|
c0c131b |
-writing to common filesystem on the server.
|
|
|
c0c131b |
-.TP 1.5i
|
|
|
c0c131b |
-.I nosharecache
|
|
|
c0c131b |
-As of kernel 2.6.18, it is no longer possible to mount the same
|
|
|
c0c131b |
-same filesystem with different mount options to a new mountpoint.
|
|
|
c0c131b |
-It was deemed unsafe to do so, since cached data cannot be shared
|
|
|
c0c131b |
-between the two mountpoints. In consequence, files or directories
|
|
|
c0c131b |
-that were common to both mountpoint subtrees could often be seen to
|
|
|
c0c131b |
-be out of sync following an update.
|
|
|
c0c131b |
-.br
|
|
|
c0c131b |
-This option allows administrators to select the pre-2.6.18 behaviour,
|
|
|
c0c131b |
-permitting the same filesystem to be mounted with different mount
|
|
|
c0c131b |
-options.
|
|
|
c0c131b |
-.br
|
|
|
c0c131b |
-.B Beware:
|
|
|
c0c131b |
-Use of this option is not recommended unless you are certain that there
|
|
|
c0c131b |
-are no hard links or subtrees of this mountpoint that are mounted
|
|
|
c0c131b |
-elsewhere.
|
|
|
c0c131b |
-.P
|
|
|
c0c131b |
-All of the non-value options have corresponding nooption forms.
|
|
|
c0c131b |
-For example, nointr means don't allow file operations to be
|
|
|
c0c131b |
-interrupted.
|
|
|
cf6d5cf |
+In the presence of multiple client network interfaces,
|
|
|
cf6d5cf |
+special routing policies,
|
|
|
cf6d5cf |
+or atypical network topologies,
|
|
|
cf6d5cf |
+the exact address to use for callbacks may be nontrivial to determine.
|
|
|
cf6d5cf |
+.SH EXAMPLES
|
|
|
cf6d5cf |
+To mount an export using NFS version 2,
|
|
|
cf6d5cf |
+use the
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+file system type and specify the
|
|
|
cf6d5cf |
+.B nfsvers=2
|
|
|
cf6d5cf |
+mount option.
|
|
|
cf6d5cf |
+To mount using NFS version 3,
|
|
|
cf6d5cf |
+use the
|
|
|
cf6d5cf |
+.B nfs
|
|
|
cf6d5cf |
+file system type and specify the
|
|
|
cf6d5cf |
+.B nfsvers=3
|
|
|
cf6d5cf |
+mount option.
|
|
|
cf6d5cf |
+To mount using NFS version 4,
|
|
|
cf6d5cf |
+use the
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file system type.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B nfsvers
|
|
|
cf6d5cf |
+mount option is not supported for the
|
|
|
cf6d5cf |
+.B nfs4
|
|
|
cf6d5cf |
+file system type.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The following example from an
|
|
|
cf6d5cf |
+.I /etc/fstab
|
|
|
cf6d5cf |
+file causes the mount command to negotiate
|
|
|
cf6d5cf |
+reasonable defaults for NFS behavior.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+.NF
|
|
|
cf6d5cf |
+.TA 2.5i +0.7i +0.7i +.7i
|
|
|
cf6d5cf |
+ server:/export /mnt nfs defaults 0 0
|
|
|
cf6d5cf |
+.FI
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Here is an example from an /etc/fstab file for an NFS version 2 mount over UDP.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+.NF
|
|
|
cf6d5cf |
+.TA 2.5i +0.7i +0.7i +.7i
|
|
|
cf6d5cf |
+ server:/export /mnt nfs nfsvers=2,proto=udp 0 0
|
|
|
cf6d5cf |
+.FI
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Try this example to mount using NFS version 4 over TCP
|
|
|
cf6d5cf |
+with Kerberos 5 mutual authentication.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+.NF
|
|
|
cf6d5cf |
+.TA 2.5i +0.7i +0.7i +.7i
|
|
|
cf6d5cf |
+ server:/export /mnt nfs4 sec=krb5 0 0
|
|
|
cf6d5cf |
+.FI
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+This example can be used to mount /usr over NFS.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+.NF
|
|
|
cf6d5cf |
+.TA 2.5i +0.7i +0.7i +.7i
|
|
|
cf6d5cf |
+ server:/export /usr nfs ro,nolock,nocto,actimeo=3600 0 0
|
|
|
cf6d5cf |
+.FI
|
|
|
cf6d5cf |
+.SH "TRANSPORT METHODS"
|
|
|
cf6d5cf |
+NFS clients send requests to NFS servers via
|
|
|
cf6d5cf |
+Remote Procedure Calls, or
|
|
|
cf6d5cf |
+.IR RPCs .
|
|
|
cf6d5cf |
+The RPC client discovers remote service endpoints automatically,
|
|
|
cf6d5cf |
+handles per-request authentication,
|
|
|
cf6d5cf |
+adjusts request parameters for different byte endianness on client and server,
|
|
|
cf6d5cf |
+and retransmits requests that may have been lost by the network or server.
|
|
|
cf6d5cf |
+RPC requests and replies flow over a network transport.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+In most cases, the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command, NFS client, and NFS server
|
|
|
cf6d5cf |
+can automatically negotiate proper transport
|
|
|
cf6d5cf |
+and data transfer size settings for a mount point.
|
|
|
cf6d5cf |
+In some cases, however, it pays to specify
|
|
|
cf6d5cf |
+these settings explicitly using mount options.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Traditionally, NFS clients used the UDP transport exclusively for
|
|
|
cf6d5cf |
+transmitting requests to servers. Though its implementation is
|
|
|
cf6d5cf |
+simple, NFS over UDP has many limitations that prevent smooth
|
|
|
cf6d5cf |
+operation and good performance in some common deployment
|
|
|
cf6d5cf |
+environments. Even an insignificant packet loss rate results in the
|
|
|
cf6d5cf |
+loss of whole NFS requests; as such, retransmit timeouts are usually
|
|
|
cf6d5cf |
+in the subsecond range to allow clients to recover quickly from
|
|
|
cf6d5cf |
+dropped requests, but this can result in extraneous network traffic
|
|
|
cf6d5cf |
+and server load.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+However, UDP can be quite effective in specialized settings where
|
|
|
cf6d5cf |
+the network’s MTU is large relative to NFS’s data transfer size (such
|
|
|
cf6d5cf |
+as network environments that enable jumbo Ethernet frames). In such
|
|
|
cf6d5cf |
+environments, trimming the
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+settings so that each
|
|
|
cf6d5cf |
+NFS read or write request fits in just a few network frames (or even
|
|
|
cf6d5cf |
+in a single frame) is advised. This reduces the probability that
|
|
|
cf6d5cf |
+the loss of a single MTU-sized network frame results in the loss of
|
|
|
cf6d5cf |
+an entire large read or write request.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+TCP is the default transport protocol used for all modern NFS
|
|
|
cf6d5cf |
+implementations. It performs well in almost every conceivable
|
|
|
cf6d5cf |
+network environment and provides excellent guarantees against data
|
|
|
cf6d5cf |
+corruption caused by network unreliability. TCP is often a
|
|
|
cf6d5cf |
+requirement for mounting a server through a network firewall.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Under normal circumstances, networks drop packets much more
|
|
|
cf6d5cf |
+frequently than NFS servers drop requests. As such, an aggressive
|
|
|
cf6d5cf |
+retransmit timeout setting for NFS over TCP is unnecessary. Typical
|
|
|
cf6d5cf |
+timeout settings for NFS over TCP are between one and ten minutes.
|
|
|
cf6d5cf |
+After the client exhausts its retransmits (the value of the
|
|
|
cf6d5cf |
+.B retrans
|
|
|
cf6d5cf |
+mount option), it assumes a network partition has occurred,
|
|
|
cf6d5cf |
+and attempts to reconnect to the server on a fresh socket. Since
|
|
|
cf6d5cf |
+TCP itself makes network data transfer reliable,
|
|
|
cf6d5cf |
+.B rsize
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.B wsize
|
|
|
cf6d5cf |
+can safely be allowed to default to the largest values supported by
|
|
|
cf6d5cf |
+both client and server, independent of the network's MTU size.
|
|
|
cf6d5cf |
+.SH "DATA AND METADATA COHERENCE"
|
|
|
cf6d5cf |
+Some modern cluster file systems provide
|
|
|
cf6d5cf |
+perfect cache coherence among their clients.
|
|
|
cf6d5cf |
+Perfect cache coherence among disparate NFS clients
|
|
|
cf6d5cf |
+is expensive to achieve, especially on wide area networks.
|
|
|
cf6d5cf |
+As such, NFS settles for weaker cache coherence that
|
|
|
cf6d5cf |
+satisfies the requirements of most file sharing types. Normally,
|
|
|
cf6d5cf |
+file sharing is completely sequential:
|
|
|
cf6d5cf |
+first client A opens a file, writes something to it, then closes it;
|
|
|
cf6d5cf |
+then client B opens the same file, and reads the changes.
|
|
|
cf6d5cf |
+.DT
|
|
|
cf6d5cf |
+.SS "Close-to-open cache consistency"
|
|
|
cf6d5cf |
+When an application opens a file stored on an NFS server,
|
|
|
cf6d5cf |
+the NFS client checks that it still exists on the server
|
|
|
cf6d5cf |
+and is permitted to the opener by sending a GETATTR or ACCESS request.
|
|
|
cf6d5cf |
+When the application closes the file,
|
|
|
cf6d5cf |
+the NFS client writes back any pending changes
|
|
|
cf6d5cf |
+to the file so that the next opener can view the changes.
|
|
|
cf6d5cf |
+This also gives the NFS client an opportunity to report
|
|
|
cf6d5cf |
+any server write errors to the application
|
|
|
cf6d5cf |
+via the return code from
|
|
|
cf6d5cf |
+.BR close (2).
|
|
|
cf6d5cf |
+The behavior of checking at open time and flushing at close time
|
|
|
cf6d5cf |
+is referred to as close-to-open cache consistency.
|
|
|
cf6d5cf |
+.SS "Weak cache consistency"
|
|
|
cf6d5cf |
+There are still opportunities for a client's data cache
|
|
|
cf6d5cf |
+to contain stale data.
|
|
|
cf6d5cf |
+The NFS version 3 protocol introduced "weak cache consistency"
|
|
|
cf6d5cf |
+(also known as WCC) which provides a way of efficiently checking
|
|
|
cf6d5cf |
+a file's attributes before and after a single request.
|
|
|
cf6d5cf |
+This allows a client to help identify changes
|
|
|
cf6d5cf |
+that could have been made by other clients.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+When a client is using many concurrent operations
|
|
|
cf6d5cf |
+that update the same file at the same time
|
|
|
cf6d5cf |
+(for example, during asynchronous write behind),
|
|
|
cf6d5cf |
+it is still difficult to tell whether it was
|
|
|
cf6d5cf |
+that client's updates or some other client's updates
|
|
|
cf6d5cf |
+that altered the file.
|
|
|
cf6d5cf |
+.SS "Attribute caching"
|
|
|
cf6d5cf |
+Use the
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+mount option to achieve attribute cache coherence
|
|
|
cf6d5cf |
+among multiple clients.
|
|
|
cf6d5cf |
+Almost every file system operation checks
|
|
|
cf6d5cf |
+file attribute information.
|
|
|
cf6d5cf |
+The client keeps this information cached
|
|
|
cf6d5cf |
+for a period of time to reduce network and server load.
|
|
|
cf6d5cf |
+When
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+is in effect, a client's file attribute cache is disabled,
|
|
|
cf6d5cf |
+so each operation that needs to check a file's attributes
|
|
|
cf6d5cf |
+is forced to go back to the server.
|
|
|
cf6d5cf |
+This permits a client to see changes to a file very quickly,
|
|
|
cf6d5cf |
+at the cost of many extra network operations.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Be careful not to confuse the
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option with "no data caching."
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+mount option prevents the client from caching file metadata,
|
|
|
cf6d5cf |
+but there are still races that may result in data cache incoherence
|
|
|
cf6d5cf |
+between client and server.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The NFS protocol is not designed to support
|
|
|
cf6d5cf |
+true cluster file system cache coherence
|
|
|
cf6d5cf |
+without some type of application serialization.
|
|
|
cf6d5cf |
+If absolute cache coherence among clients is required,
|
|
|
cf6d5cf |
+applications should use file locking. Alternatively, applications
|
|
|
cf6d5cf |
+can also open their files with the O_DIRECT flag
|
|
|
cf6d5cf |
+to disable data caching entirely.
|
|
|
cf6d5cf |
+.SS "The sync mount option"
|
|
|
cf6d5cf |
+The NFS client treats the
|
|
|
cf6d5cf |
+.B sync
|
|
|
cf6d5cf |
+mount option differently than some other file systems
|
|
|
cf6d5cf |
+(refer to
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+for a description of the generic
|
|
|
cf6d5cf |
+.B sync
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.B async
|
|
|
cf6d5cf |
+mount options).
|
|
|
cf6d5cf |
+If neither
|
|
|
cf6d5cf |
+.B sync
|
|
|
cf6d5cf |
+nor
|
|
|
cf6d5cf |
+.B async
|
|
|
cf6d5cf |
+is specified (or if the
|
|
|
cf6d5cf |
+.B async
|
|
|
cf6d5cf |
+option is specified),
|
|
|
cf6d5cf |
+the NFS client delays sending application
|
|
|
cf6d5cf |
+writes to the server
|
|
|
cf6d5cf |
+until any of these events occur:
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+Memory pressure forces reclamation of system memory resources.
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+An application flushes file data explicitly with
|
|
|
cf6d5cf |
+.BR sync (2),
|
|
|
cf6d5cf |
+.BR msync (2),
|
|
|
cf6d5cf |
+or
|
|
|
cf6d5cf |
+.BR fsync (3).
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+An application closes a file with
|
|
|
cf6d5cf |
+.BR close (2).
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+The file is locked/unlocked via
|
|
|
cf6d5cf |
+.BR fcntl (2).
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+In other words, under normal circumstances,
|
|
|
cf6d5cf |
+data written by an application may not immediately appear
|
|
|
cf6d5cf |
+on the server that hosts the file.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+If the
|
|
|
cf6d5cf |
+.B sync
|
|
|
cf6d5cf |
+option is specified on a mount point,
|
|
|
cf6d5cf |
+any system call that writes data to files on that mount point
|
|
|
cf6d5cf |
+causes that data to be flushed to the server
|
|
|
cf6d5cf |
+before the system call returns control to user space.
|
|
|
cf6d5cf |
+This provides greater data cache coherence among clients,
|
|
|
cf6d5cf |
+but at a significant performance cost.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Applications can use the O_SYNC open flag to force application
|
|
|
cf6d5cf |
+writes to individual files to go to the server immediately without
|
|
|
cf6d5cf |
+the use of the
|
|
|
cf6d5cf |
+.B sync
|
|
|
cf6d5cf |
+mount option.
|
|
|
cf6d5cf |
+.SS "Using file locks with NFS"
|
|
|
cf6d5cf |
+The Network Lock Manager protocol is a separate sideband protocol
|
|
|
cf6d5cf |
+used to manage file locks in NFS version 2 and version 3.
|
|
|
cf6d5cf |
+To support lock recovery after a client or server reboot,
|
|
|
cf6d5cf |
+a second sideband protocol --
|
|
|
cf6d5cf |
+known as the Network Status Manager protocol --
|
|
|
cf6d5cf |
+is also required.
|
|
|
cf6d5cf |
+In NFS version 4,
|
|
|
cf6d5cf |
+file locking is supported directly in the main NFS protocol,
|
|
|
cf6d5cf |
+and the NLM and NSM sideband protocols are not used.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+In most cases, NLM and NSM services are started automatically,
|
|
|
cf6d5cf |
+and no extra configuration is required.
|
|
|
cf6d5cf |
+Configure all NFS clients with fully-qualified domain names
|
|
|
cf6d5cf |
+to ensure that NFS servers can find clients to notify them of server reboots.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+NLM supports advisory file locks only.
|
|
|
cf6d5cf |
+To lock NFS files, use
|
|
|
cf6d5cf |
+.BR fcntl (2)
|
|
|
cf6d5cf |
+with the F_GETLK and F_SETLK commands.
|
|
|
cf6d5cf |
+The NFS client converts file locks obtained via
|
|
|
cf6d5cf |
+.BR flock (2)
|
|
|
cf6d5cf |
+to advisory locks.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+When mounting servers that do not support the NLM protocol,
|
|
|
cf6d5cf |
+or when mounting an NFS server through a firewall
|
|
|
cf6d5cf |
+that blocks the NLM service port,
|
|
|
cf6d5cf |
+specify the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+mount option. NLM locking must be disabled with the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+option when using NFS to mount
|
|
|
cf6d5cf |
+.I /var
|
|
|
cf6d5cf |
+because
|
|
|
cf6d5cf |
+.I /var
|
|
|
cf6d5cf |
+contains files used by the NLM implementation on Linux.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Specifying the
|
|
|
cf6d5cf |
+.B nolock
|
|
|
cf6d5cf |
+option may also be advised to improve the performance
|
|
|
cf6d5cf |
+of a proprietary application which runs on a single client
|
|
|
cf6d5cf |
+and uses file locks extensively.
|
|
|
cf6d5cf |
+.SS "NFS version 4 caching features"
|
|
|
cf6d5cf |
+The data and metadata caching behavior of NFS version 4
|
|
|
cf6d5cf |
+clients is similar to that of earlier versions.
|
|
|
cf6d5cf |
+However, NFS version 4 adds two features that improve
|
|
|
cf6d5cf |
+cache behavior:
|
|
|
cf6d5cf |
+.I change attributes
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.IR "file delegation" .
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.I change attribute
|
|
|
cf6d5cf |
+is a new part of NFS file and directory metadata
|
|
|
cf6d5cf |
+which tracks data changes.
|
|
|
cf6d5cf |
+It replaces the use of a file's modification
|
|
|
cf6d5cf |
+and change time stamps
|
|
|
cf6d5cf |
+as a way for clients to validate the content
|
|
|
cf6d5cf |
+of their caches.
|
|
|
cf6d5cf |
+Change attributes are independent of the time stamp
|
|
|
cf6d5cf |
+resolution on either the server or client, however.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+A
|
|
|
cf6d5cf |
+.I file delegation
|
|
|
cf6d5cf |
+is a contract between an NFS version 4 client
|
|
|
cf6d5cf |
+and server that allows the client to treat a file temporarily
|
|
|
cf6d5cf |
+as if no other client is accessing it.
|
|
|
cf6d5cf |
+The server promises to notify the client (via a callback request) if another client
|
|
|
cf6d5cf |
+attempts to access that file.
|
|
|
cf6d5cf |
+Once a file has been delegated to a client, the client can
|
|
|
cf6d5cf |
+cache that file's data and metadata aggressively without
|
|
|
cf6d5cf |
+contacting the server.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+File delegations come in two flavors:
|
|
|
cf6d5cf |
+.I read
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.IR write .
|
|
|
cf6d5cf |
+A
|
|
|
cf6d5cf |
+.I read
|
|
|
cf6d5cf |
+delegation means that the server notifies the client
|
|
|
cf6d5cf |
+about any other clients that want to write to the file.
|
|
|
cf6d5cf |
+A
|
|
|
cf6d5cf |
+.I write
|
|
|
cf6d5cf |
+delegation means that the client gets notified about
|
|
|
cf6d5cf |
+either read or write accessors.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Servers grant file delegations when a file is opened,
|
|
|
cf6d5cf |
+and can recall delegations at any time when another
|
|
|
cf6d5cf |
+client wants access to the file that conflicts with
|
|
|
cf6d5cf |
+any delegations already granted.
|
|
|
cf6d5cf |
+Delegations on directories are not supported.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+In order to support delegation callback, the server
|
|
|
cf6d5cf |
+checks the network return path to the client during
|
|
|
cf6d5cf |
+the client's initial contact with the server.
|
|
|
cf6d5cf |
+If contact with the client cannot be established,
|
|
|
cf6d5cf |
+the server simply does not grant any delegations to
|
|
|
cf6d5cf |
+that client.
|
|
|
cf6d5cf |
+.SH "SECURITY CONSIDERATIONS"
|
|
|
cf6d5cf |
+NFS servers control access to file data,
|
|
|
cf6d5cf |
+but they depend on their RPC implementation
|
|
|
cf6d5cf |
+to provide authentication of NFS requests.
|
|
|
cf6d5cf |
+Traditional NFS access control mimics
|
|
|
cf6d5cf |
+the standard mode bit access control provided in local file systems.
|
|
|
cf6d5cf |
+Traditional RPC authentication uses a number
|
|
|
cf6d5cf |
+to represent each user
|
|
|
cf6d5cf |
+(usually the user's own uid),
|
|
|
cf6d5cf |
+a number to represent the user's group (the user's gid),
|
|
|
cf6d5cf |
+and a set of up to 16 auxiliary group numbers
|
|
|
cf6d5cf |
+to represent other groups of which the user may be a member.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Typically, file data and user ID values appear unencrypted
|
|
|
cf6d5cf |
+(i.e. "in the clear") on the network.
|
|
|
cf6d5cf |
+Moreover, NFS versions 2 and 3 use
|
|
|
cf6d5cf |
+separate sideband protocols for mounting,
|
|
|
cf6d5cf |
+locking and unlocking files,
|
|
|
cf6d5cf |
+and reporting system status of clients and servers.
|
|
|
cf6d5cf |
+These auxiliary protocols use no authentication.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+In addition to combining these sideband protocols with the main NFS protocol,
|
|
|
cf6d5cf |
+NFS version 4 introduces more advanced forms of access control,
|
|
|
cf6d5cf |
+authentication, and in-transit data protection.
|
|
|
cf6d5cf |
+The NFS version 4 specification mandates NFSv4 ACLs,
|
|
|
cf6d5cf |
+RPCGSS authentication, and RPCGSS security flavors
|
|
|
cf6d5cf |
+that provide per-RPC integrity checking and encryption.
|
|
|
cf6d5cf |
+Because NFS version 4 combines the
|
|
|
cf6d5cf |
+function of the sideband protocols into the main NFS protocol,
|
|
|
cf6d5cf |
+the new security features apply to all NFS version 4 operations
|
|
|
cf6d5cf |
+including mounting, file locking, and so on.
|
|
|
cf6d5cf |
+RPCGSS authentication can also be used with NFS versions 2 and 3,
|
|
|
cf6d5cf |
+but does not protect their sideband protocols.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B sec
|
|
|
cf6d5cf |
+mount option specifies the RPCGSS security mode
|
|
|
cf6d5cf |
+that is in effect on a given NFS mount point.
|
|
|
cf6d5cf |
+Specifying
|
|
|
cf6d5cf |
+.B sec=krb5
|
|
|
cf6d5cf |
+provides cryptographic proof of a user's identity in each RPC request.
|
|
|
cf6d5cf |
+This provides strong verification of the identity of users
|
|
|
cf6d5cf |
+accessing data on the server.
|
|
|
cf6d5cf |
+Note that additional configuration besides adding this mount option
|
|
|
cf6d5cf |
+is required in order to enable Kerberos security.
|
|
|
cf6d5cf |
+Refer to the
|
|
|
cf6d5cf |
+.BR rpc.gssd (8)
|
|
|
cf6d5cf |
+man page for details.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Two additional flavors of Kerberos security are supported:
|
|
|
cf6d5cf |
+.B krb5i
|
|
|
cf6d5cf |
+and
|
|
|
cf6d5cf |
+.BR krb5p .
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B krb5i
|
|
|
cf6d5cf |
+security flavor provides a cryptographically strong guarantee
|
|
|
cf6d5cf |
+that the data in each RPC request has not been tampered with.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B krb5p
|
|
|
cf6d5cf |
+security flavor encrypts every RPC request
|
|
|
cf6d5cf |
+to prevent data exposure during network transit; however,
|
|
|
cf6d5cf |
+expect some performance impact
|
|
|
cf6d5cf |
+when using integrity checking or encryption.
|
|
|
cf6d5cf |
+Similar support for other forms of cryptographic security (such as lipkey and SPKM3)
|
|
|
cf6d5cf |
+is also available.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The NFS version 4 protocol allows
|
|
|
cf6d5cf |
+clients and servers to negotiate among multiple security flavors
|
|
|
cf6d5cf |
+during mount processing.
|
|
|
cf6d5cf |
+However, Linux does not yet implement such negotiation.
|
|
|
cf6d5cf |
+The Linux client specifies a single security flavor at mount time
|
|
|
cf6d5cf |
+which remains in effect for the lifetime of the mount.
|
|
|
cf6d5cf |
+If the server does not support this flavor,
|
|
|
cf6d5cf |
+the initial mount request is rejected by the server.
|
|
|
cf6d5cf |
+.SS "Mounting through a firewall"
|
|
|
cf6d5cf |
+A firewall may reside between an NFS client and server,
|
|
|
cf6d5cf |
+or the client or server may block some of its own ports via IP
|
|
|
cf6d5cf |
+filter rules.
|
|
|
cf6d5cf |
+It is still possible to mount an NFS server through a firewall,
|
|
|
cf6d5cf |
+though some of the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command's automatic service endpoint discovery mechanisms may not work; this
|
|
|
cf6d5cf |
+requires you to provide specific endpoint details via NFS mount options.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+NFS servers normally run a portmapper or rpcbind daemon to advertise
|
|
|
cf6d5cf |
+their service endpoints to clients. Clients use the rpcbind daemon to determine:
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+What network port each RPC-based service is using
|
|
|
cf6d5cf |
+.IP
|
|
|
cf6d5cf |
+What transport protocols each RPC-based service supports
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The rpcbind daemon uses a well-known port number (111) to help clients find a service endpoint.
|
|
|
cf6d5cf |
+Although NFS often uses a standard port number (2049),
|
|
|
cf6d5cf |
+auxiliary services such as the NLM service can choose
|
|
|
cf6d5cf |
+any unused port number at random.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Common firewall configurations block the well-known rpcbind port.
|
|
|
cf6d5cf |
+In the absense of an rpcbind service,
|
|
|
cf6d5cf |
+the server administrator fixes the port number
|
|
|
cf6d5cf |
+of NFS-related services so that the firewall
|
|
|
cf6d5cf |
+can allow access to specific NFS service ports.
|
|
|
cf6d5cf |
+Client administrators then specify the port number
|
|
|
cf6d5cf |
+for the mountd service via the
|
|
|
cf6d5cf |
+.BR mount (8)
|
|
|
cf6d5cf |
+command's
|
|
|
cf6d5cf |
+.B mountport
|
|
|
cf6d5cf |
+option.
|
|
|
cf6d5cf |
+It may also be necessary to enforce the use of TCP or UDP
|
|
|
cf6d5cf |
+if the firewall blocks one of those transports.
|
|
|
cf6d5cf |
+.SS "NFS Access Control Lists"
|
|
|
cf6d5cf |
+Solaris allows NFS version 3 clients direct access
|
|
|
cf6d5cf |
+to POSIX Access Control Lists stored in its local file systems.
|
|
|
cf6d5cf |
+This proprietary sideband protocol, known as NFSACL,
|
|
|
cf6d5cf |
+provides richer access control than mode bits.
|
|
|
cf6d5cf |
+Linux implements this protocol
|
|
|
cf6d5cf |
+for compatibility with the Solaris NFS implementation.
|
|
|
cf6d5cf |
+The NFSACL protocol never became a standard part
|
|
|
cf6d5cf |
+of the NFS version 3 specification, however.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The NFS version 4 specification mandates a new version
|
|
|
cf6d5cf |
+of Access Control Lists that are semantically richer than POSIX ACLs.
|
|
|
cf6d5cf |
+NFS version 4 ACLs are not fully compatible with POSIX ACLs; as such,
|
|
|
cf6d5cf |
+some translation between the two is required
|
|
|
cf6d5cf |
+in an environment that mixes POSIX ACLs and NFS version 4.
|
|
|
cf6d5cf |
.SH FILES
|
|
|
cf6d5cf |
+.TP 1.5i
|
|
|
cf6d5cf |
.I /etc/fstab
|
|
|
cf6d5cf |
-.SH "SEE ALSO"
|
|
|
cf6d5cf |
-.BR fstab "(5), " mount "(8), " umount "(8), " exports (5)
|
|
|
cf6d5cf |
-.SH AUTHOR
|
|
|
cf6d5cf |
-"Rick Sladkey" <jrs@world.std.com>
|
|
|
cf6d5cf |
+file system table
|
|
|
cf6d5cf |
.SH BUGS
|
|
|
cf6d5cf |
+The generic
|
|
|
cf6d5cf |
+.B remount
|
|
|
cf6d5cf |
+option is not fully supported.
|
|
|
cf6d5cf |
+Generic options, such as
|
|
|
cf6d5cf |
+.BR rw " and " ro
|
|
|
cf6d5cf |
+can be modified using the
|
|
|
cf6d5cf |
+.B remount
|
|
|
cf6d5cf |
+option,
|
|
|
cf6d5cf |
+but NFS-specific options are not all supported.
|
|
|
cf6d5cf |
+The underlying transport or NFS version
|
|
|
cf6d5cf |
+cannot be changed by a remount, for example.
|
|
|
cf6d5cf |
+Performing a remount on an NFS file system mounted with the
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option may have unintended consequences.
|
|
|
cf6d5cf |
+The
|
|
|
cf6d5cf |
+.B noac
|
|
|
cf6d5cf |
+option is a mixture of a generic option,
|
|
|
cf6d5cf |
+.BR sync ,
|
|
|
cf6d5cf |
+and an NFS-specific option
|
|
|
cf6d5cf |
+.BR actimeo=0 .
|
|
|
cf6d5cf |
.P
|
|
|
cf6d5cf |
-Checking files on NFS filesystem referenced by file descriptors (i.e. the
|
|
|
cf6d5cf |
-.BR fcntl
|
|
|
cf6d5cf |
-and
|
|
|
cf6d5cf |
-.BR ioctl
|
|
|
cf6d5cf |
-families of functions) may lead to inconsistent result due to the lack of
|
|
|
cf6d5cf |
-consistency check in kernel even if noac is used.
|
|
|
cf6d5cf |
+Before 2.4.7, the Linux NFS client did not support NFS over TCP.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Before 2.4.20, the Linux NFS client used a heuristic
|
|
|
cf6d5cf |
+to determine whether cached file data was still valid
|
|
|
cf6d5cf |
+rather than using the standard close-to-open cache coherency method
|
|
|
cf6d5cf |
+described above.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Starting with 2.4.22, the Linux NFS client employs
|
|
|
cf6d5cf |
+a Van Jacobsen-based RTT estimator to determine retransmit
|
|
|
cf6d5cf |
+timeout values when using NFS over UDP.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Before 2.6.0, the Linux NFS client did not support NFS version 4.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+Before 2.6.8, the Linux NFS client used only synchronous reads and writes
|
|
|
cf6d5cf |
+when the
|
|
|
cf6d5cf |
+.BR rsize " and " wsize
|
|
|
cf6d5cf |
+settings were smaller than the system's page size.
|
|
|
cf6d5cf |
+.P
|
|
|
cf6d5cf |
+The Linux NFS client does not yet support
|
|
|
cf6d5cf |
+certain optional features of the NFS version 4 protocol,
|
|
|
cf6d5cf |
+such as security negotiation, server referrals, and named attributes.
|
|
|
cf6d5cf |
+.SH "SEE ALSO"
|
|
|
cf6d5cf |
+.BR fstab (5),
|
|
|
cf6d5cf |
+.BR mount (8),
|
|
|
cf6d5cf |
+.BR umount (8),
|
|
|
cf6d5cf |
+.BR mount.nfs (5),
|
|
|
cf6d5cf |
+.BR umount.nfs (5),
|
|
|
cf6d5cf |
+.BR exports (5),
|
|
|
cf6d5cf |
+.BR nfsd (8),
|
|
|
cf6d5cf |
+.BR rpc.idmapd (8),
|
|
|
cf6d5cf |
+.BR rpc.gssd (8),
|
|
|
cf6d5cf |
+.BR rpc.svcgssd (8),
|
|
|
cf6d5cf |
+.BR kerberos (1)
|
|
|
cf6d5cf |
+.sp
|
|
|
cf6d5cf |
+RFC 768 for the UDP specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 793 for the TCP specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 1094 for the NFS version 2 specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 1813 for the NFS version 3 specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 1832 for the XDR specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 1833 for the RPC bind specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 2203 for the RPCSEC GSS API protocol specification.
|
|
|
cf6d5cf |
+.br
|
|
|
cf6d5cf |
+RFC 3530 for the NFS version 4 specification.
|