Potential History file corruption?

[Alex Kiernan]

Russ Allbery <rra at stanford.edu> writes:

> Alex Kiernan <alexk at demon.net> writes:
> > Russ Allbery <rra at stanford.edu> writes:
> 
> >> We're separately having all kinds of problems with BerkeleyDB with
> >> Cyrus's IMAP server.  :/ I'm really starting to believe that the
> >> problems with ovdb are actually BerkeleyDB problems.
> 
> > Thats *very* interesting... are you seeing similar kinds of symptoms?
> 
> What we've been getting is that, after a couple of days of running,
> everything that tries to access the database just hangs.  Including the
> program that's supposed to break deadlocks.

Sounds very similar, its likely stuck in __db_mutex_lock/unlock.

> I don't remember enough of the threads here to remember if that's
> also one of the various symptoms people are seeing.
> 

If Cyrus is good about logging errors, there's probably an EINVAL
returned somewhere from one of the ->put calls just before it locks
solid.

> Some kind of locking problem somewhere, I'm sure, but we've not tracked it
> down yet.
> 

If you can build with Purify, it'd be interesting to know if you're
seeing the same kind of uninitialised reads as I'm seeing.

-- 
Alex Kiernan, Principal Engineer, Development, Thus PLC