[OpenAFS] Redhat 9 hangs during backups when there is local client activity

Joe Buehler jbuehler@hekimian.com
Mon, 23 Aug 2004 08:40:53 -0400


I am having a problem with a couple Redhat 9 machines hanging
during AFS backups.  The problem appears to be triggered by
cronjobs that use the AFS client on the fileserver that is
backing itself up.

My initial fix has been to move the cronjobs to avoid the backup
window.  One of the machines crashed (not a hang) this
weekend however.  The oops on the console says that there was
an invalid page fault in afs_readdir() (name approximate)
that was due to a "find" process.  Which means a cronjob.

Our fileservers are running 1.2.10 compiled to use LWP instead
of the broken Redhat 9 NPTL (what a mess that was...).

I want to ask, is this a known issue that has been fixed?

If someone wants to try it, it's pretty easy to duplicate -- do
a backup (we dump to disk files which then get saved by Legato)
at the same time that a find command is running under /afs,
both on the same machine.
-- 
Joe Buehler