I can trigger a "crash" by writing a 'c' to /proc/sysrq-trigger. I am using this to simulate a kernel panic and check the system's behaviour after a panic (we want it to reboot).
The reason I am wondering about the difference is that sysctl is showing "panic=0", which means don't reboot after a panic. To enable a reboot, a numerical value (indicating seconds of delay) should be written to this value.
So given that the server does reboot after triggering a crash, what's the difference between a crash and a panic?
Is there a difference - is it that a crash can happen where a kernel error is not caught, so there is no opportunity to call panic() ?