SUMMARY: Problems with Automounter and rexd

Date: Tue May 05 1992 - 00:34:16 CDT

Hi Everyone:

I wanted to wait until I actually fixed the problem using the suggestions
given. Currently I'm awaiting the suggested patch from Sun. However I
got several requests as to what I findings were, so I figured I might as well
post this summary without actually implementing the fix.

Also I received answers sating that its okay to kill the automounter
on a client machine, (with the kill -TERM option). I've done so in the
past on Suns, RS/6000, DECs, and HPs and there is no apparent problem.

However folks say never to try killing the automounter on a server.
I didn't specify this, but my query was in regard to a server. I got my

My thanks to: (Leslie Dreyer +1 215 770 2795) I got Leslie's response
                                                even before I received a copy
                                                of my own posting :-) (Bill Rea, Universityof Canterbury, New Zealand) (Jim Davis) (Kevin W. Thomas) (Brent Callaghan) Eckhard R"uggeberg


-Vispi Dumasia

In my original message I wrote: =============================== Hi I'm having a strange problem. Maybe some of you might have experienced/seen it. I have a Sparc 4/370 server running SunOS.4.1.1 32M mem, 100M swap. NIS, Sun's automounter, NFS and DNS running.

The problem I'm seeing is (and I can't duplicate it at will ....) the machine hangs because the automounter screws up. I get messages like (/home@pid1234) NFS server antares not responding. Once this happens only a reboot will fix it.

I have seen this happen only when some user uses the "on" command. Since the "on" command forks off a rexd process, I believe that the automounter and the rexd process both write to the /etc/mtab file and overwrite/erase entries n the .etc.mtab file in some way.

I occasionally get messages on the console that say. automount: WARNING. illegal entry in line 1873 in /etc/mtab file

If this might be a lead. Sometime there is more than one automount process running on the machine at the same time. And once I also happened to see one automount process with a fixed (unchanging) pid, and another automount process resulting in a different pid every time I did a ps -auwwx | grep automount I guess the automounter went haywire and started forking several automount processes which would die instantaneously.

Is there any patches, fixes for this problem. Any clues would be very helpful. Also I curious, can one kill the automount process and restart it without rebooting the machine, and expect consistent results.

Many thanks.

-Vispi Dumasia (713) 579 4871 ===============================

SUMMARY: ======== I got several responses to it. All of them said the same thing.

Get the automount jumbo patch for 4.1 (which also needs to be applied to 4.1.1) This patch is based on 4.1.2 source.

The README to the patch follows:

Patch-ID# 100249-03 Keywords: automount symbolic link timeout, jumbo Synopsis: SunOS 4.1;4.1.1;4.1.2: automounter JUMBO patch Date: 3/Mar/92 SunOS release: 4.1 4.1.1 4.1.2 Bug ID's fixed with this patch: 1038308, 1048875, 1044048, 1083225 Patches which conflict with this patch: Architectures for which this patch is available: sun3, sun4, *sun3x, *sun4c *sun3 binary can be used for sun3x, sun4 binary can be used for sun4c

Problem description: Bug 1083225 (March 3, 1992, -03) -------------------------------- This version of automount is based on SunOS4.1.2 Source. If an entry is removed from /etc/mtab while the automounter daemon is reading it, it is possible for the automounter to get into a state where it will loop forever, printing "WARNING: /etc/mtab: line <number>: bad entry" on the console and hang the server. Bug 1038308: In 4.1, the automounter sets up a different symbolic link 5 min. after the first access. This happens if there are multiple locations for the same mount in the auto mount map entry, and those locations don't share the same subdir.

Bug 1048875 The automounter forks due to rpc.rexd. This problem is usually seen when using the on command. The automounter will complain it cannot find /etc/mtab. Bug 1044048 Automounter consumes a steady, increasing percentage of memory. Coinciding with this is the consumption of swap space as well. INSTALL: As root: mv /usr/etc/automount /usr/etc/automount.fcs cp automount.`arch` /usr/etc/automount chown root /usr/etc/automount chgrp staff /usr/etc/automount chmod 755 /usr/etc/automount


PS: Sun takes 3 days to send the patch to you by email. 5 days by magnetic media. ==========================

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:42 CDT