[OpenAFS-devel] Re: DB server failover on master?

[Andrew Deason]

On Thu, 29 Dec 2011 18:33:57 -0500 (EST)
Benjamin Kaduk <kaduk@MIT.EDU> wrote:

> I got a kernel core from a previous hang (a 'cp' process), and there
> wasn't anything that looked like it was going to deadlock; nobody held
> the glock, either.

If it helps... if it's the same rx conn/call each time, look at the
values in the conn and call structure to see why it's not timing out. If
it's a different call each time, see why the afs_server structure isn't
getting marked down after the network failure (are none of the vl
servers marked down in the core?)

-- 
Andrew Deason
adeason@sinenomine.net