I'm running Autofs (v4) on a RH9 system with a 2.6.0-test1 kernel, and I'm seeing some
problems:
1) mounts that happen via automount never seem to time out
2) forcing an umount fails ('device is busy')
Are you still seeing "expired <mount>" messages in your syslog even though they're not umounting?
I'm seeing similar behavior, but I can get them to umount by hand. We have a huge direct map here that we're mounting via a program that uses awk to get matches from the map, mount if one match is found, or mount a submount process if multiple matches are found.
Recently I've noticed more hosts with a ton of mounts that aren't in use (and the load on the host always rises to whatever the number of cpus is). I run a little script that does a mount -t nfs, does a quick fuser, and if not in use, umounts it... and they free up with the load immediately dropping down. I also then see a bunch of messages like:
Jul 22 15:16:23 shockley automount[5175]: automount.c:767: assertion failed: ap.state == ST_READY
Jul 22 15:16:23 shockley automount[27731]: expired /usr/local/projects/vlsi/<blah>
Jul 22 15:16:23 shockley automount[5175]: shutting down, path = /usr/local/projects/vlsi/<blah>
Jul 22 15:16:23 shockley automount[12691]: automount.c:767: assertion failed: ap.state == ST_READY
but everything seems to be ok after about a minute as all the remaining submount processes clean themselves up and stop.
-- Mike Marion-Unix SysAdmin/Staff Engineer-http://www.miguelito.org Homer: "Wow! $350?!? Now I can buy 70 transcripts of Nightline!" ==> Simpsons
_______________________________________________ autofs mailing list [EMAIL PROTECTED] http://linux.kernel.org/mailman/listinfo/autofs
