automount — automatic server mount / unmount daemon
automount [-V] [-d] [-D type] [-1] [-tm secs] [-tl secs] [-s] [-tcp]
[-m directory map -mnt directory] …
automount is a daemon that automatically mounts network filesystems when
they are first accessed and later unmounts them when they are idle.
automount creates a virtual filesystem mounted at one or more places in
the client’s file and directory hierarchy. Potential server mount points
within this virtual filesystem appear as symbolic links. Reading a sym-
bolic link causes automount to mount the associated remote filesystem in
a separate hierarchy and to return the path to the real mount point as
the target of the symlink.
To make the « trigger » symbolic links used by automount distinguishable
from normal symbolic links, the sticky bit is set in the mode flags for
the link. Programs which would normally traverse symbolic links can test
for this bit and avoid triggering the mount. Various parts of the sys-
tem, including ls(1), have been modified in this way.
Each virtual filesystem created by automount is governed by a correspond-
ing map. One or more maps and the location in the hierarchy where they
are to appear may be specified on the command line with -m and -mnt:
-m directory map
is followed by a specification of the private mount directory where the
actual mounts are to be established:
Each map’s hierarchy is rooted at the directory specified.
A map may be file or a special map. A file map is a file containing a
list of entries of the form:
location mount_options server:path
mount_options must be a comma-separated list of options drawn from the
options known to mount(8) and mount_nfs(8). automount will automatically
make available at the map’s mount point the directory specified by
server:path (as a symlink). When the symlink is traversed, it will mount
the server in the directory specified with -mnt and return the real mount
point as the result of the symlink resolution.
In addition to reading files specifying mount maps, automount supports
the « -fstab », « -static », and « -nsl » maps. The « -fstab » and
« -static » maps are derived from fstab(5) data (as provided by
All mounts in fstab(5) without the « net » option will be made to appear
at the fstab(5)-specified location in the form of a symlink into the
directory where the « -static » map is actually mounted (itself a direc-
tory of symlinks). For example, the entry:
server:/Network/Applications /Network/Applications nfs nosuid 0 0
and automount invocation
automount -m /automount/static -static -mnt /private/var/automount
result in a non-trigger symlink /Network/Applications pointing to /auto-
mount/static/Network/Applications which, when traversed, causes
server:/Network/Applications to be mounted on /private/var/automount/Net-
work/Applications (which would then be returned from reading the link).
All mounts with the « net » option will be mounted within the « -fstab »
map’s filesystem using a path of the form:
For such mounts, the path specified in the fstab(5) entry will be
ignored. For example, if the fstab(5) database contained an entry for
and automount was started as follows:
automount -m /Useful -fstab -mnt /private/Useful
then a symlink would appear in /Useful, /Useful/polaris/Library/Fonts,
and link to /private/Useful/polaris/Library/Fonts where the filesystem is
actually mounted when the symlink is traversed.
The « -nsl » map generates its filesystem hierarchy from information gen-
erated by NSL. NSL « neighborhoods » are presented as directories, popu-
lated by server entries.
Accessing servers in the NSL hierarchy may prompt the user for authenti-
-V Print version and host information, then quit
-m directory map
Associate the specified map with the given directory. The direc-
tory will be created if it doesn’t exist. map may be the name of
a file, or it may be the name of a special map. See the FILE
MAPS and SPECIAL MAPS sections above.
-d Run automount in debug mode. The program remains in the fore-
ground and sends debugging information to standard output.
Log debug messages for type. type may be « mount », « proc »,
« mount », « select », « options », « nsl », or « all ». Multiple -D
options may be specified.
-1 (The numeric digit « one ».) Modifies the « -fstab » and
« -static » maps to do mounts « one at a time », when an actual
mount point is traversed rather than mounting all mounts from a
given server when the first of its mounts is referenced and
Set the timeout for NFS mounts to secs seconds. The default
value is 20 seconds. The « mnttimeo=n » mount option overrides
Set the time-to-live for NFS mounts to secs seconds. The default
value is 3600 seconds. The « ttl=n » mount option overrides this
automount periodically checks all its mounted filesystems. If it
finds any filesystems that have been idle for their associated
time-to-live value, it will attempt to unmount them. An unmount
will only be successful if there are no processes with open files
in that filesystem. Unless -1 is specified, if one or more
mounts from a given server are found to be active, any mounts
from that server that were successfully unmounted will be immea-
Supplying a ttl value of 0 will disable this behavior and allow
servers to remain mounted forever.
-s Force all mounts at startup and never expire any mounts.
-tcp Mount servers using TCP if possible, otherwise using UDP (the
default is to try UDP first, then TCP). Per mount_nfs(8), « -T »,
« TCP », or « tcp » mount options have the same effect as specifying
-tcp; « -U », « UDP », or « udp » mount options force the default
behavior of trying UDP first even if -tcp is specified.