We are trying to troubleshoot an issue with random system lockups during medium - heavy loads on an Ubuntu MySQL 5.1 server.
There is nothing logging into the system during these lockups. The only thing logging is to the console "task hung for 120 seconds" on task kjournald, mysqld, etc. when a load spike hits and the system totally crashes. There is nothing even in the dmesg log.
Another strange thing (which might or might not be related to teh above) is happening during system load is time drift. The load on the system is proportionate to the time dift. I have tried deleting /var/lib/ntp/ntp.drift but it keeps coming back to 500.00 . Kernel version is 2.6.38-15-server , Ubuntu 10.04, running under a Xen 3.4 kernel/host.