On 02/21/2013 01:19 PM, Quanah Gibson-Mount wrote:
--On Thursday, February 21, 2013 9:27 AM -0800 Quanah Gibson-Mount quanah@zimbra.com wrote:
--On Thursday, February 21, 2013 8:35 AM -0500 "Robert W. Smith" bob.smith@o3bnetworks.com wrote:
Hi,
I am running a 4-way multi-master configuration with a number of slaves in remote locations. I am currently running openldap 2.4.33 on top of CentOS 6.3 (I built 2.4.33 from a modified base centos 6 spec file). I was originally running the centos base openldap 2.4.23 using N-way multimaster using the syncrepl configuration but I was having problems with the masters and slaves staying in perfect sync--other than this 2.4.23 was running stably since last spring. I'll try to be brief in what has happened since Feb 1.
Did you build openldap with debugging symbols? (-g) Did you disable optimization? (-O0)
If so, I would advise submitting an ITS, with a full backtrace from gdb:
Oh, also make sure when make install is run, you don't strip the binaries of the debug symbols (make install STRIP="")
I would also warn against using the centos spec file, as it links to NSS instead of OpenSSL. You probably want to look at http://ltb-project.org/wiki/download#openldap
--Quanah
--
Quanah Gibson-Mount Sr. Member of Technical Staff Zimbra, Inc A Division of VMware, Inc.
Zimbra :: the leader in open source messaging and collaboration
Quanah,
I re-compiled 2.4.33 using the openssl libraries on Friday and have installed the updated packages on my four master servers. I have not experienced a single deadlock since then. The slaves have not been upgraded yet--they are still running the centos stock 2.4.23 distribution.
The only issue that I still have is one or more threads running at full throttle and utilizing 100% CPU. But this is a different issue and I'll start a new thread on this when I get a chance to look at this more.
Thanks, Bob --bs