[OpenAFS-devel] Re: CopyOnWrite failure

[hoffman@cs.pitt.edu]

Hello,

I've been following this thread in the archive and figured I'd
better join in.

To answer Derrick's questions:

> Some questions for those of you with this problem:
> -only with linux servers?

Yes, but we're running OpenAFS only on Linux at this point.  All of
our other AFS servers are running Transarc AFS.  We had the volume
corruption problem with Transarc AFS on Linux, too, if that's any clue.

> -always with non-replicated volumes that have a .backup?

Correct, but we have only one volume that's both replicated and
backed up daily and it's resident on a Digital Unix fileserver.

> -if the above, was the backup being recreated at the time? (the VolserLog
>  may be helpful here, as well as the vos examine info)

It seems to happen at the first access by the user after the
recloning.  Vos examine says that it can't attach the volume.

> -what if anything pertinent about access patterns?

Nonen that I can tell.

> -can you try a lwp fileserver for some period of time on your hardware?

I did that and it hasn't helped.  In fact, it may be worse.

Most recently, I applied Ted Anderson's patch to vol/ihandle.c.
I'll let you know if that helps any.

Thanks to all for your help.

	---Bob.
--
Bob Hoffman, N3CVL      University of Pittsburgh           Tel: +1 412 624 8490
hoffman@cs.pitt.edu     Department of Computer Science     Fax: +1 412 624 8854