Miner, Jonathan W (CSC) (US SSA) wrote:
From: fedora-list-bounces(a)redhat.com on behalf of Miner, Jonathan W (CSC) (US SSA)
Sent: Thu 07/26/2007 09:51 AM
Subject: Kernel 188.8.131.52-27.fc7 breaks NFS
Just upgraded from 2.6.21-1.3228.fc7, and now seeing problems with NFS. I originally
believed the problem was with the automounter, but it looks like it might be NFS instead.
Rebooting with the old kernel solves the problem.
In this example, I'll create a new mountpoint, and try to mount my home directory.
The directory is being served from a SUN Cluster 2.1 server.
# rm -rf /h
# mkdir -p /h/jminer
# mount superfly:/superfly/vol02/home/jminer /h/jminer
mount.nfs: /h/jminer is already mounted or busy
# echo $?
Now, if I try to mount the directory from the physical node, the mount suceeds.
# mount fly_a:/superfly/vol02/home/jminer /h/jminer
# echo $?
I saw similar problems with an earlier FC release, but it was related to either the
autofs or nfs-utils package.
I just learned something else. If I stop the autofs (/etc/init.d/autofs stop), then the
original mount command works fine. If autofs is restarted, the mount command fails again.
If you don't need automount don't run it. If you do, please post your
config file. I have seen similar when a lower directory is mounted, or
in the automount config, like mounting superfly:/superfly/vol02/home as
Some portion of problems with new kernels result from added error
checking in the new kernel telling you that you have been doing it wrong
and getting away with it. I'm not saying that happened here, but this
sounds a lot as if automount has some part of that mount in config, or
being used, or otherwise referenced.
You might try running automount, check that the mount fails, then run
lsof and see if automount is using something related.
Bill Davidsen <davidsen(a)tmr.com>
"We have more to fear from the bungling of the incompetent than from
the machinations of the wicked." - from Slashdot