Performing some test for new software support for redundant network interfaces, we have testers down in the heavy labs pulling out and plugging in network cables. One such scenerio now triggers a reboot, with the very last thing before reboot in /var/log/messages showing a message from one the device driver for the interface:
Oct 3 14:17:45 coralm227 kernel: [1875109.684268] mlx4_en: eth5: Close port called # ... reboot in this interval Oct 3 14:44:48 coralm227 syslog-ng: syslog-ng starting up; version='2.0.9'
My best guess is that the device driver is triggering a kernel crash, and the system is rebooting. However, I see no oops messages in /var/log/messages (nor anywhere else in /var/log/ that I can find), and see no signs of a kernel core dump anywhere?
How can I tell if the system is configured to dump core on kernel trap, and if it is, where would I find the core dump path? If the system is not configured to dump core, how can this be done? Are there any other log files that are relevant for tracking down the source of the reboot? Any other tips for debugging this auto-reboot scenerio?