On 06/11/98 12:33:52 you wrote:
This problem seems to happen about once a month. Wondering if it's a configuration probably on my end or just an issue with my filer.
Mail is delivered to ~/.mailbox onto a F540 running 4.3.1D7. and clients are all Solaris 2.5.1.
After about a month I start getting lots of calls from users (mostly elm users, some mutt users) about being unable to read their mailbox. A truss on the process shows it sleeping waiting for a flock on .mailbox. mutt uses dotlocking and seems to just hang doing an open().
Only solution seems to be a reboot. Any clues or pointers on where I should be looking?
Reboot the filer or the Solaris client? This wasn't clear.
There is a good chance that the bug lies in the Solaris lockd, not in the filer. You should apply all of the NFS and locking patches for 2.5.1 (heack, all of the recommended ones really) and see if it goes away. If it doesn't or you can't do this, as an interim measure killing and restarting the locking process on the UNIX client may be a less drastic solution, although I suppose under the right conditions this may result in corrupted files elsewhere in the filesystem.
A lock management utility on the client side to browse through and remove bad locks, or a similar command on the filer, would be very useful.
Bruce