Re: SUMMARY: How do you umount NFS mounts of hosts that are down?

From: Sheryl Coppenger (sheryl@eesun.gwu.edu)
Date: Fri Jun 28 1991 - 13:35:01 CDT


I know the mailing list isn't the place for discussions, so please
e-mail me any further comments and I will post a summary if warranted.

I noticed in the posting referenced in the Subject line, that
several people who responded implied that the automounter would
take care of everything in situations where an NFS server went down.
That hasn't been true in our case, and I thought I would bring
up some of our recent problems while NFS was fresh in everyone's minds.

We had the case where a disk died on an NFS server. The disk was
replaced and the contents restored from tape. To be sure, we had
"stale NFS file handle" messages instead of hangs. But umount and mount
had no effect. Editing mtab, restarting nfsd and biod also had no
effect. On the directory where the mount points were located, an
"ls" command showed the mount point and an "ls -l" gave the "stale
NFS file handle" message. I was unable to alter the mount point
in any way. This was the case on about half the machines, generally
those which had users logged on whose home directories were on the
server. Logging off the users and killing processes did not seem to
kick automount free. Since the NFS directories could not be remounted
when the machine came up, the fact that we had an immediate return
rather than a hang wasn't much consolation.

We finally wound up rebooting all of the affected machines. Our
network is made up of a mixture of HPs and Suns. The HPs don't have
automounter, all of the Suns do. The behavior was almost identical
except that all HPs were affected and the Suns who were using the
disk were.

We have witnessed another annoying behavior from the automounter.
We have had cron scripts which failed when they tried to run executables
which were located on an NFS directory that should have been automounted.
Also, make files running overnight have gone off into space, apparently
because the directory they were using was unmounted. Has anyone seen
this happen before? Is there a workaround/patch? We are running
4.1 on Sun3s and 4.1.1 on Sun4s with no patches installed.

Thanks for any replies.

-- 

Sheryl Coppenger SEAS Computing Facility Staff sheryl@seas.gwu.edu The George Washington University (202) 994-6853



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:21 CDT