Hello. I'm having problems with the automounter and RPC timeouts.
Note that this is an effect of a bad networking situation, but such
situations can happen, and the result is hung NFS mount points because
the automounter won't stop generating an error. Here is the problem
I'm having.
I have a bunch of automounted NFS directories for some of our users.
They will run remote jobs on machines, which automount their home
directories.
Sometimes, the network goes hokey, and we get the following types of errors:
Sep 16 15:46:49 node002 automount[493]: attempting to mount entry /home/clim
Sep 16 15:47:09 node002 automount[8208]: >> mount: RPC: Timed out
Sep 16 15:47:09 node002 automount[8208]: mount(nfs): nfs: mount failure
emcnfs410:/mnt_1003/ucode/clim on /home/clim
Sep 16 15:47:09 node002 automount[493]: attempting to mount entry /home/clim
Sep 16 15:47:29 node002 automount[8414]: >> mount: RPC: Timed out
Sep 16 15:47:29 node002 automount[8414]: mount(nfs): nfs: mount failure
emcnfs410:/mnt_1003/ucode/clim on /home/clim
Sep 16 15:47:29 node002 automount[493]: attempting to mount entry /home/clim
Sep 16 15:47:49 node002 automount[8627]: >> mount: RPC: Timed out
Sep 16 15:47:49 node002 automount[8627]: mount(nfs): nfs: mount failure
emcnfs410:/mnt_1003/ucode/clim on /home/clim
etc. Note that these can keep going for a while if the network dies
or gets badly screwed up.
Now, once the network comes back to normal, I will sometimes get the
following errors when trying to access the bad directory (in this case,
/home/clim):
Sep 16 18:17:54 node002 kernel: autofs warning: lookup failure on positive dentry,
status = -4, name = clim
And this *never* goes away on it's own. For example, I do a "df", and
it hangs at the /home/clim mount point. I can CTRL-C it, and try
again, and it keeps generating these errors.
Now, I *can* fix it by hand. If I first shut down the automounter,
and then unmount everything by hand (including the self-generated
auto-mount points sometimes, such as /emc in this case, that suddenly
appear in "df") I can then restart the automounter, and it will then
start working just fine again.
But, if I don't disable the automounter first, and just try to
unmount, it hangs indefinitely with these "autofs warnings" forever.
Any ideas on why this is happening, anything I can try, etc? I am
going to try to create a test case, instead of relying on a flaky
network to generate these tests for me. But, this problem is
currently killing me whenever we have any sort of networking
problem...
Thanks very much!!!
Note that I am currently running 2.2.10, with *many* NFS client
patches applied so that the NFS client is working quite well. It's
just the automounter giving me fits now. :)
-- Steve McClure
[EMAIL PROTECTED]
