Re: RE24 testing call (2.4.27 official call #1)
Quanah Gibson-Mount wrote:
--On Tuesday, November 08, 2011 12:13 PM -0800 Quanah Gibson-Mount
<quanah(a)zimbra.com> wrote:
> --On Tuesday, November 08, 2011 8:27 PM +0100 Michael Ströder
> <michael(a)stroeder.com> wrote:
>
>> Quanah Gibson-Mount wrote:
>>> Please test RE24. Thanks!
>>
>> There is something fishy in test044. It seg faults every now and then.
>> Sometimes after 20-30 iterations, sometimes sooner. Not sure whether it's
>> only due to back-mdb.
>>
>> ITS with details will follow. Others should bang on this as well.
>
> 50 iterations completed for me no issue. I'm trying again with 500
> iterations. I assume you are seeing this specifically with the back-mdb
> backend.
>
> I'm using ./run -l 500 -b mdb test044
Zero failures for me after over 1000 runs. OTOH, Howard's been able to
reproduce it twice and is looking at how to fix it. ;)
Stack overrun in back-mdb. back_mdb_search is a bit more memory hungry than
back_bdb_search, unfortunately. The nested search call that dynlist invokes is
enough to put it over the top.
--
-- Howard Chu
CTO, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/