summary: automounter

From: Benson I. Margulies (
Date: Fri Dec 27 1991 - 23:41:15 CST

In two messages, I asked:

1) why does the automounter put things in /tmp_mnt? How do people

2) why does the automounter create circular links on the machine
with the local mount of the file system?

3) is AMD any better?

In brief:

1) It always puts things in /tmp_mnt (or wherever -M is) to avoid
interposing itself in all NFS requests. People cope by depending
on shells that remember what you typed, or by replacing /bin/pwd
with a program that strips /tmp_mnt.

2) If you are going to let the real home machine of a file system see
an automounter map for that file system, you have to mount it in a
different place on the home machine than the automounter mounts it.
Then the automounter's propensity for symbolic links will create
a link at the auto-mount-point to the actual mount point. Of course,
this defeats the /tmp_mnt stripping program above.

(this reminds me of the old joke about the mathematician who, having
successfully extinguished a house on fire, ignited the next house he
was presented with so as to reduce it to a problem with a known

2) in less brief terms

My original goal was to use the automounter as a simple way to
distribute the fstab. I have an existing collection of mounts,
generally in the root. I hoped to avoid having to edit 20 fstabs every
time a new file system was added to a server.

The SunOS automounter is unsuitable to this task. I'm not willing to
replace such a basic utility as /bin/pwd, nor to goose all the tools
that look at its output, nor to teach 40 people not to complain that
pwd is giving them crazy results.

>From the responses I got, I concluded that if you were willing to
design a set of file system, local mount points, and remote mount
points with the automounter in mind, it might indeed be useful for
something. I'm not sure I see what, but I'm probably just dense.

For those who wondered what I was going to do, I'm going to write
a script that runs perl via rsh to edit fstabs semi-automatically.

The respondents were too numerous to thank individually.

happy new year


This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:22 CDT