[Pacemaker] Corosync 2.3 dies randomly

[Robert Parsons]

I'm trying to build out a web farm cluster using Corosync/Pacemaker. I 
started with the stock versions in Ubuntu 12.04 but did not have a lot 
of success. I removed the corosync (1.x) and pacemaker packages and 
built Corosync 2.3 and Pacemaker 1.1.9 from source. It generally seems 
to run better but I am having big issues with Corosync. I have 14 
completely identical nodes. They differ only in ip address and host 
name. Periodically, corosync will fail to start up on boot. It's not 
consistent and happens randomly. What's even worse, once the cluster is 
up Corosync will occasionally die for no apparent reason. There are not 
errors logged. Nothing. The process simply disappears, taking the node 
offline.

My cluster has zero stability thanks to this Corosync issue. Anyone got 
any ideas?

Thanks.

- Rob P.