Server instance will "hang" within minutes or hours while undergoing heavy loads of concurrent sessions and requests.
There is no more response seen from any type of client request.
Nothing is written to our /logs directory since the server froze so abruptly that disk i/o was also halted.
There is no heavy use of memory during this time, so no Out Of Memory issue are suspected.
The server's java process will not respond to a kill -3, so no thread dump is possible.
Killing the 2 java pids with a -9 option will kill the processes and allow you to start up the server again.
However, if you have a JDK installed, running a Jstack (heap dump) against the child java pid does free up the "hang" and allows the server to suddenly resume running normally (no restart needed).
Issue/Introduction
This article discusses a Red hat bug that may affect the use of Tibco Data Virtualization.
Resolution
This has been reported as a Red Hat bug in RH v6.6, 7.0 and 7.1, confirmed to be fixed in v6.9.
Verify this Red Hat bug by comparing the thread stack from your server to the one listed in the Red Hat bug report.
Look on your server host in this path: /proc/<parent pid>/task/<child pid>/stack Run a 'more' cmd on the 'stack' file and see if it looks like this: