On Tue, 05 Dec 2006, Aaron Richton might have said:
Take a core snapshot of the process with gcore (or just attach with gdb, if you don't need to get things running quickly) and get a symbolic backtrace. On FC5, you probably need -debuginfo to get useful backtraces.
I've reported at least one livelock since 2.3.19 (ITS #4738). That would show up as 100% CPU, if it hit you. An update has been released for about a month; upgrade OpenLDAP to 2.3.30.
On Tue, 5 Dec 2006, Mike wrote:
Slapd was fine this weekend, then yesterday it went to 100% of the CPU and load average of 1.0+. No filesystems are full and I can find no mention of an issue in /var/log/*. Where can I look next for what's going on?
I had this happen a few weeks ago, but I thought it was due to a full file system at the same time.
Mike
Red Hat Fedora Core 5 with all the latest patches [mikee@host ~]$ rpm -qa | grep ldap openldap-devel-2.3.19-4 ldapjdk-4.17-1jpp_3fc.1.1 openldap-clients-2.3.19-4 nss_ldap-249-1 python-ldap-2.0.6-5.2.1 cyrus-sasl-ldap-2.1.21-10 openldap-servers-2.3.19-4 smbldap-tools-0.9.2-3.fc5 openldap-2.3.19-4 mod_authz_ldap-0.26-6.2.1
I did 'service ldap restart' and the CPU times are back to normal. I also did a 'yum check-config' and there is no update yet from the FC5 server.
Thanks.
Mike