Nothing like this has been mentioned in the forum before. Any clues in the message log? Updating is probably a good idea, in particular since the UEK is a tickless kernel. I would also suggest to check the virtual environment and VM software, and consider that the client's computer could be the source of the problem too.
Random lock-ups: oh, joy.
A server typically becomes autistic when it experiences either memory fragmentation, memory starvation, or a storage I/O failure that congests the system.
Oracle has an "OSWatcher" tool, that is really just some shell scripts that run top(1), vmstat(8), free(1) and the like on a regular basis. Every server should always be running OSWatcher; it's cheap insurance.
You should also setup the KDUMP facility to dump core to a LOCAL DISK in the event of a kernel panic.
The system console is an invaluable source of diagnostic information, always keep the moral equivalent of a terminal attached to it.
- When the lockup occurs, can you SSH in to the server?
- During the lockup, can you "root" login at the console?
- How do you recover from the lock-up? Command "# reboot"? System reset? Power cycle? The answer to this question helps determine just how locked-up the server is.