[OpenAFS] Errors: Fileserver freezes, Volumes contains orphans

[Rubino Geiß]

> From: openafs-info-admin@openafs.org 
> [mailto:openafs-info-admin@openafs.org] On Behalf Of Derrick 
> J Brashear
> On Thu, 30 Jan 2003, [iso-8859-1] Rubino Geiß wrote:
> 
> > Hi,
> >
> > Klaas Hagemann reported some problems with dying 
> fileservers. I’ve the 
> > sad duty to report another, jet different incident.
> 
> Has this still been happening? Have you considered the "make 
> pthreaded processes dump core" kernel patch I posted about?
> 
> Anyone else?
> 
> It would be good if we could get this fixed before OpenAFS 
> 1.2.9 is released, but we need pretty much anything to go on. 

I'm sorry, but we haven’t had another incident – maybe I shouldn’t be sorry,
but happy ;) 

As I posted some time ago, it’s very unlikely to happen anytime soon again.
It only happens to one of our servers and only 2 times within the past 11
months. We do not know how to reproduce it.

During the next service cycle we will take measurements to get cores the
next time.


ATTENTION! In my original posting I have reported 2 maybe unconnected
problems. One of our homedir volumes contains infrequently many orphans!
That’s why the ”Volumes contains orphans” is in the message subject. At the
last incident we discovered more volumes containing orphans on the failing
server, but unrelated(?) to that one certain volumes (that lives on an other
server) contains orphans, too.

Up to now I haven’t got any explanation for this. Hint: Last spring this
volume suffered from CoW bug, but we salvaged it replaced the fileserver
with a new one and everything seems fine, but know orphans again! Is the
some strange permanence?

Maybe someone can elaborate on how volumes get orphans?

Thanks & Bye, Ruby