[OpenAFS-devel] Stability problems, using Linux 2.4.14 or higher and openafs

Tim C. tim@umbc.edu
Mon, 24 Jun 2002 21:56:52 -0400


> using vanilla linux 2.4.14 and linux 2.4.18 together with openafs-1.2.3,
> openafs-1.2.4 and openafs-1.2.5, we do have serious problems with the
> long term stability of our openafs clients. After some days or weeks of
> uptime we end up on all of our PIII SMP boxes with a situation where
> /afs is still mounted but cannot be accessed anymore. No kernel messages
> can be found in the syslog files and the only way to solve this
> situation is to reboot the machine.
>
> Hmm.....,does anybody experience similar problems. Maybe our problem is
> related to the deadlock problem on linux-2.4 discussed in this list
> about 3 weeks ago?
>
  We've got three multi-login machines running Redhat 7.2, with a custom kernel
2.4.17(with xfs and devfs) on a dual P3 machine with openafs 1.2.2 built from
source.  And we haven't expirienced any problem like that before(in any
version/hw).  The three machines each usually have ~ 40 users on them at a
time.  One of those has been up for about 100 days, and behaves fine.  We're
not using nfs on the machines.
  Also we have workstations running stock redhat 7.3, with openafs-1.2.4
without any problems.

  As for your problem, is there anything on the console?  Maybe something about
loosing contact with a server or something?  I've seen afs behave that way when
the machine looses contact with the servers, usually due to network or routes.

I hope this helps some.
  Tim

-----------------------------------------------------------------------
Tim Craig		These are my opinions and not my employers. :)
OIT-Systems	&	Imaging Research Center
tim@umbc.edu		It's hard to be serious when you're
			naked. - Garfield
-----------------------------------------------------------------------