[Date Prev][Date Next]
RE: back-bdb deadlocks
All of a sudden the lock detector is working on my SuSE 6 system again,
but the test takes far too long. With no transactions the test runs in
just over a minute. With transactions it takes about 35 minutes on my
system. I added a ldap_pvt_thread_yield() after the txn_abort() and that
brought the runtime down to 23 minutes. Finally I got sick of it and just
decided to serialize the whole thing. As an experiment, I obtained a BDB
lock on the parent DN. That brings the test down to 3 minutes, 0 deadlocks.
I made this change in bdb_add and bdb_delete but bdb_modrdn will need it as
well. Of course this works because the current test has all the adds and
deletes occurring in the same branch of the DIT, I'll have to test again with
add/deletes occurring elsewhere in the tree as well.
-- Howard Chu
Chief Architect, Symas Corp. Director, Highland Sun
Symas: Premier OpenSource Development and Support
> -----Original Message-----
> From: Kurt D. Zeilenga [mailto:Kurt@OpenLDAP.org]
> Sent: Thursday, January 17, 2002 6:09 AM
> To: Howard Chu
> Cc: openldap-devel@OpenLDAP.org
> Subject: RE: back-bdb deadlocks
> At 09:52 PM 2002-01-16, Howard Chu wrote:
> >I still get locked up running the updated test script. Also the
> number of locks
> >in use keeps growing up. Looks like a lock leak. (Say that 5 times
> fast...) I'm
> >surprised you're not seeing the same thing.
> It ran all night here without a problem.
> I do have one problem, seems I cannot run db_ utilities concurrently
> with slapd and, if I do, slapd does hang. This may be a local
> configuration problem. I'm going to re-configure to remove that
> as a problem source and retest, hopefully later today.