I have similar problem on my two nodes cluster used as GFS2 storage cluster. Just GFS2 not service clustering. Both nodes were hanging several times and we had to reboot . They are running webservers. I have been in contact with Redhat and I was advised to usegfs2_hangalyzer during the next IO or process hang. The Redhat support engineer is assuming it is a lock contention issue, and on your case you said only your developers can solve this type of problem. What do you mean by that ? Can you make it a little bit clear for me. You mean the problem is caused by applications running in your servers ? In our case we have Java and Tomcat are running, you mean the Java developers should see the matter ?
By the way I just subscribed to thelinux-cluster redhat com now, is this correct way to post messages ?
Very wel, now i make a deep analysis, but i think thath the problemi s simply the conteention, so only our developers can solve the problem.