The occsd.bin process is the core process which belongs to the Cluster Synchronization Service. This service is responsible for node membership decisions, so when it no longer exists, the server is rebooted. I was attempting to better understand the library calls made by the process, so I issued an ltrace command against the pid for ocssd.bin. About four seconds later the node was rebooted with the ltrace command returning…
[root@expressdb2 ~]# ltrace -p 2011
+++ killed by SIGTRAP +++
[root@expressdb2 ~]#
The moral of the story is, don’t issue ltrace against ocssd.bin 🙂
UPDATE 2014/07/03:
This is apparently resolved in RHEL 6.3
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/6.3_Technical_Notes/ltrace.html