[Ltb-users] segfault slapd

Ulrich.Tehrani at t-systems.com Ulrich.Tehrani at t-systems.com
Mon May 8 19:26:23 CEST 2017


Hi Clement,

Master and  replica are using both:

openldap-ltb-2.4.44-2.el6.x86_64
berkeleydb-ltb-4.6.21.NC-4.el6.patch4.x86_64
cyrus-sasl-2.1.23-15.el6_6.2.x86_64

I use Red Hat Enterprise Linux Server release 6.8 (Santiago)  as a VM on KVM

Before  I installed the LTB  packages I run 2.4.40  from the Redhat Distribution.  Here the behavior was similar to the current behaviour
Because  the crashes always occurred on the replica I was convinced that is a problem with syncrepl.
Redhat didn’t deliver a newer package than 2.4.40 so I decided to switch to the LTB package 2.4.44  last week.

But I still have the problems with crashing replica.  I never got  a core although I set ulimit –c unlimited

I will try your suggestion with gdb. Maybe this will help.


Regards

Uli





Von: ltb-users [mailto:ltb-users-bounces at lists.ltb-project.org] Im Auftrag von Clément OUDOT
Gesendet: Montag, 8. Mai 2017 18:13
An: ltb-users at lists.ltb-project.org
Betreff: Re: [Ltb-users] segfault slapd




Le 08/05/2017 à 15:25, Ulrich.Tehrani at t-systems.com<mailto:Ulrich.Tehrani at t-systems.com> a écrit
Hi all,

After  one week without problem my ltb-slapd  replica crashed  last night with:

May  8 03:48:45 uvairz0022 rsyslogd-2177: imuxsock lost 87 messages from pid 30032 due to rate-limiting
May  8 03:48:56 uvairz0022 kernel: slapd[30106]: segfault at 8 ip 0000000000520c9d sp 00007f270f67b160 error 4
May  8 03:48:56 uvairz0022 kernel: slapd[31063]: segfault at 8 ip 0000000000520c9d sp 00007f2707e7c0d0 error 4 in slapd[400000+287000] in slapd[400000+287000]
May  8 03:48:56 uvairz0022 kernel:

This is the only info I have.  As a workaround I implemented an automatic restart if a  ldapsearch has no access.
What can I do to really solve the problem?

I saw  there Is also a LTB debug package.  Is there any configuration required for the debug package?
Will the debug package provide more infos if slapd will crash again?


Hello Ulrich,

can you tell us which package do you use (RPM/Deb) on which OS?

Installing debug package will let you the possibility to run slapd into gdb and to get a stacktrace when segfault happens. See also http://www.openldap.org/faq/data/cache/59.html


--

Clément OUDOT

Consultant en logiciels libres, Expert infrastructure et sécurité

Savoir-faire Linux

137 boulevard de Magenta - 75010 PARIS

Blog: http://sflx.ca/coudot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ltb-project.org/pipermail/ltb-users/attachments/20170508/e954e05e/attachment-0001.html>


More information about the ltb-users mailing list