Robert Moskowitz wrote:

I have a test system: Centos 5.2 on an OQO, that has been hanging hard. I have to unplug it and pull the battery so I can then cold start it.

This last time all I did was open a terminal window and SU to root, then start the lastest build of SIP Communicator (which uses JRE 1.6.0_10). I was not even making a test phone call at the time. Oh, and the system only runs IPv6, no v4 addressing.

So how do I find out what is causing the hard lockups?

if the problem is triggered reproducibly by building SIP C, then just don't use X when starting the build; change to a console (Ctrl-Alt-F1) and watch out for kernel messages indicating a software (kernel) problem.

If the problem is due to hardware, there is a chance that memtest86+ will find bad memory. If it's a bad disk, you can check with smartctl. There's are burn-in programs available; I use burnCPU to put load on multi-core systems. Compiling the Linux kernel a hundred times in a row is also a good test; not a single build should fail.

HTH,

Kay