The
utility
is a daemon that automatically mounts file systems
whenever a file or directory
within that file system is accessed.
File systems are automatically unmounted when they
appear to be quiescent.
The
utility
operates by attaching itself as an
NFS
server to each of the specified
directories
Lookups within the specified directories
are handled by
,
which uses the map defined by
mapname
to determine how to resolve the lookup.
Generally, this will be a host name, some file system information
and some mount options for the given file system.
In the first form depicted above,
will print a short help string.
In the second form, if no options are
specified, or if the
-F
is used,
will read configuration parameters from the file
conf_file
which defaults to
/etc/amd.conf
The last form is described below.
OPTIONS
-a temporary-directory
Specify an alternative location for the real mount points.
The default is
/.amd_mnt
-c duration
Specify a
duration
in seconds, that a looked up name remains
cached when not in use.
The default is 5 minutes.
-d domain
Specify the local domain name.
If this option is not
given the domain name is determined from the hostname.
-k kernel-arch
Specifies the kernel architecture.
This is used solely
to set the ${karch} selector.
-l logfile
Specify a logfile in which to record mount and unmount events.
If
logfile
is the string
``syslog
''
then the log messages will be sent to the system log daemon by
syslog(3).
The default syslog facility used is
LOG_DAEMON
If you
wish to change it, append its name to the log file name, delimited by a
single colon.
For example, if
logfile
is the string
``syslog:local7
''
then
will log messages via
syslog(3)
using the
LOG_LOCAL7
facility (if it exists on the system).
-n
Normalize hostnames.
The name referred to by ${rhost} is normalized relative to the
host database before being used.
The effect is to translate
aliases into
``official''
names.
-o op_sys_ver
Override the compiled-in version number of the operating system.
Useful
when the built in version is not desired for backward compatibility reasons.
For example, if the build in version is
``2.5.1
''
you can override it to
``5.5.1
''
and use older maps that were written with the latter in mind.
-p
Print PID.
Outputs the process ID of
to standard output where it can be saved into a file.
-r
Restart existing mounts.
The
utility
will scan the mount file table to determine which file systems
are currently mounted.
Whenever one of these would have
been auto-mounted,
inherits
it.
-t timeout.retransmit
Specify the NFS
timeout
interval,
in tenths of a second, between
NFS/RPC
retries (for UDP only).
The default
is 0.8 seconds.
The second value alters the retransmit counter, which
defaults to 11 retransmissions.
Both of these values are used by the kernel
to communicate with amd.
Useful defaults are supplied if either or both
values are missing.
The
amd(8)
utility
relies on the kernel RPC retransmit mechanism to trigger mount retries.
The values of these parameters change the overall retry interval.
Too long
an interval gives poor interactive response; too short an interval causes
excessive retries.
-v
Version.
Displays version and configuration information on standard error.
-w interval
Specify an
interval
in seconds, between attempts to dismount
file systems that have exceeded their cached times.
The default is 2 minutes.
-x options
Specify run-time logging options.
The options are a comma separated
list chosen from:
fatal , error , user , warn , info , map , stats , all
-y domain
Specify an alternative
NIS
domain from which to fetch the
NIS
maps.
The default is the system domain name.
This option is ignored if
NIS
support is not available.
-A arch
Specifies the OS architecture.
This is used solely to set the ${arch}
selector.
-C cluster-name
Specify an alternative HP-UX cluster name to use.
-D option
Select from a variety of debug options.
Prefixing an
option with the string
``no
''
reverses the effect of that option.
Options are cumulative.
The most useful option is
``all
''
Since
-D
is only used for debugging other options are not documented here:
the current supported set of options is listed by the
-v
option
and a fuller description is available in the program source.
-F conf_file
Specify an
configuration file to use.
See
amd.conf5
for description of this file's format.
This configuration file is used to
specify any options in lieu of typing many of them on the command line.
The
amd.conf5
file includes directives for every command line option amd has, and many
more that are only available via the configuration file facility.
The
configuration file specified by this option is processed after all other
options had been processed, regardless of the actual location of this option
on the command line.
-H
Print help and usage string.
-O op_sys_name
Override the compiled-in name of the operating system.
Useful when the
built in name is not desired for backward compatibility reasons.
For
example, if the build in name is
``sunos5
''
you can override it to
``sos5
''
and use older maps which were written with the latter in mind.
-S
Do not lock the running executable pages of
into memory.
To improve
's
performance, systems that support the
plock(3)
call, could lock the
process into memory.
This way there is less chance
the operating system will schedule, page out, and swap the
process as
needed.
This tends to improve
's
performance, at the cost of reserving the
memory used by the
process (making it unavailable for other processes).
If this behavior is not desired, use the
-S
option.
-T tag
Specify a tag to use with
amd.conf5.
All map entries tagged with
tag
will be processed.
Map entries that are not tagged are always processed.
Map entries that are tagged with a tag other than
tag
will not be processed.
FILES
/.amd_mnt
directory under which file systems are dynamically mounted
/etc/amd.conf
default configuration file
CAVEATS
Some care may be required when creating a mount map.
Symbolic links on an
NFS
file system can be incredibly inefficient.
In most implementations of
NFS
their interpolations are not cached
by the kernel and each time a symbolic link is encountered during a
lookuppn
translation it costs an
RPC
call to the
NFS
server.
It would appear that a large improvement in real-time
performance could be gained by adding a cache somewhere.
Replacing symlinks with a suitable incarnation of the auto-mounter
results in a large real-time speedup, but also causes a large
number of process context switches.
A weird imagination is most useful to gain full advantage of all
the features.
