Re: 2.2.13 crashes with ch_realloc of 2455080002 bytes failed

[Quanah Gibson-Mount <quanah@stanford.edu>]

--On Thursday, June 17, 2004 6:52 PM +0300 Janis Pinkis <jpinkis@lanet.lv> 
wrote:

> Citçju Quanah Gibson-Mount <quanah@stanford.edu>:
>
> > --On Thursday, June 17, 2004 4:52 PM +0300 Janis Pinkis
> > <jpinkis@lanet.lv>  wrote:
> >
> > >
> > > It happened, when i was making a massive data import, with ldapadd and
> > > in paralel rstarted a consumer server. Both hosts are runnig RHEL3 and
> > > openldap-2.2.13.
>
> > This is either a problem with RedHat, or your configuration of RedHat.
> > Make sure that slapd is started with ulimit -n.
> >
> > You also did not include your DB_CONFIG file.
> >
> > Please see my post from yesterday:
> >
> > <http://www.openldap.org/lists/openldap-software/200406/msg00472.html>
>
> By the way, i tried to restart consumer once more, and my provider slapd
> crashed again without any messages in syslog. But it happens, only, if
> i'am doing ldapadd on provider, if not, i can restart the consumer how i
> like, nothing bad happens. One more thing, both machines are SMP
> machines, with 4 CPUs each.

I suggest using slurpd instead of syncRepl.  I'm still seeing issues around 
using syncRepl in my development environment that keep me from considering 
it for use in our production environment.

Your DB_CONFIG looks alright, except I tend to only set this:

#
# Set database flags.
# (for database loading/reindexing)
set_flags       DB_TXN_NOSYNC

if I'm doing a slapadd (*not* ldapadd).

Again, the memory issue you are seeing appears to be directly related to 
RedHat, as the several people who've reported it were all using RedHat, and 
I can't duplicate it under Debian.  I may have a RedHat box to test this on 
@ Stanford in the near future.

--Quanah

--
Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html