Quanah Gibson-Mount wrote:

--On Monday, March 09, 2009 3:24 PM +0000 michael@stroeder.com wrote:

...

michael@stroeder.com wrote:

...

Full_Name: Michael Str?der Version: RE24

Seems to work in HEAD on the very same system.

RE24 is not ready for testing yet.

Well, Howard recently recommended to use it (see below). So I'm testing as usual...

Ciao, Michael.

-------- Original Message -------- Subject: Re: RE24 (2.4.15) slapd consuming large amounts of CPU Date: Sat, 07 Mar 2009 00:47:57 -0800 From: Howard Chu hyc@symas.com To: John Morrissey jwm@horde.net CC: openldap-software@openldap.org References: 20090304185521.GA3184@boost.horde.net 49AEE757.3000502@symas.com 20090306141020.GA7876@boost.horde.net

John Morrissey wrote:

On Wed, Mar 04, 2009 at 12:40:55PM -0800, Howard Chu wrote:

...

John Morrissey wrote:

...

A backtrace (below) shows many threads waiting on a mutex in bdb_cache_*().

From the trace it looks similar to ITS#5860. The patches for this are in HEAD, not yet released.

With an shm_key, slapd CPU consumption dropped noticably and has been running long enough to make the leak apparent.

Since this second-round fix for ITS#5860 has been merged into RE24, do you recommend running that, or should I cherry-pick 1.120.2.{21,22} into our local packaging?

At this point I would just use RE24. Unless any new critical bugs show up in the next few days, it's very likely what will be 2.4.16.