Re: seg faults with slapd and libc?

Good info; will look into LTB,

Upgraded to centos 6.7; same issue / segfault.

Lots of CPU's / memory; HPC world:

ie:

[root@afiles]# cat /proc/cpuinfo | grep proc | wc

64 192 950

[root@afiles]#

[root@afiles]# free -g

total used free shared buffers cached

Mem: 125 10 115 0 0 1

-/+ buffers/cache: 8 117

Swap: 31 0 31

Versions:

[root@afiles]# rpm -qa| grep openldap

openldap-devel-2.4.40-5.el6.x86_64

openldap-2.4.40-5.el6.x86_64

openldap-clients-2.4.40-5.el6.x86_64

openldap-servers-2.4.40-5.el6.x86_64

[root@afiles]#

I am proxying two ldap servers into one; starting to think this is the issue. Moved the internal bdp off the proxy LDAP server and running stress tests now using ldapperf.

#LOCAL MGMT2 DATABASE

#database bdb

#cachesize 50000

#idlcachesize 50000

#suffix "dc=test,dc=com"

#checkpoint 1024 15

#rootdn "cn=test,dc=test,dc=com"

#rootpw

#directory /var/lib/ldap

#Indices to maintain for this database

#index objectClass eq,pres

#index ou,cn,mail,surname,givenname eq,pres,sub

#index uidNumber,gidNumber,loginShell eq,pres

#index uid,memberUid eq,pres,sub

#index nisMapName,nisMapEntry eq,pres,sub

#database meta - COMBINES the LDAP DATABASES

database meta

suffix "dc=local"

rootdn "cn=test,dc=local"

rootpw aaaa

#internel LDAP

#uri "ldap://127.0.0.1/ou=internal,dc=local"

#lastmod off

#suffixmassage "ou=internal,dc=local" "dc=test,dc=com"

#external - customer LDAP

#uncomment lines and only change vars inside [] to match env

#uri "ldap://[myldap]/ou=external,dc=local"

#lastmod off

#suffixmassage "ou=external,dc=local" "[dc=datacenter,dc=company,dc=com]"

##external ldap

uri "ldap://sertve/ou=external,dc=local"

lastmod off

suffixmassage "ou=external,dc=local" "dc=datacenter,dc=company,dc=com"

Jason K Cafarelli
Desk: (508) 637-5705 (primary)
Mobile: (508) 215-9712
jason.cae@gmail.com

On Sat, Sep 19, 2015 at 2:00 PM, Michael Ströder <michael@stroeder.com> wrote:

jason cafarelli wrote:
> centos 6.5 with a basic slapd setup; internal LDAP and proxy to external
> LDAP

Which OpenLDAP version is this?

Also note that RedHat heavily patches OpenLDAP and links it against libnss.
You should therefore also open a ticket for RHEL.

> Getting segaults every 6 hours or so when I let QA on system and generate
> some typical user load.
>
> All 32 threads halt; ldapsearch stops responding.

Do you have 8 CPU cores?

> Typically no logs; last one dropped this:
>
> ie:
>
> Sep 18 07:47:58 maryann2 kernel: slapd[53754]: segfault at 7f3083ffff98 ip
> 00007f3104cc3532 sp 00007f30a5ff9660 error 6 in libc-2.12.so
> [7f3104c4a000+18b000]
>
>
> Should i just sledgehammer a yum update and get off centos 6.5?

The openldap-technical is not a RedHat forum. But being in your position I'd
upgrade within the RHEL 6.x series first.

Also consider using the LTB project's OpenLDAP RPMs. Those are linked to
OpenSSL libs and are more close to upstream OpenLDAP releases.

http://ltb-project.org/wiki/documentation/openldap-rpm

Ciao, Michael.