If you’ve set up your VM Linux server to produce a crash dump directory/file if the server crashes, and it doesn’t seem to work check the following:
By default, kdump will use /var/crash – make sure this is big enough. If it’s not you may find a message in /var/log/boot.log to that effect.
Despite this, today I still couldn’t get the crash dump file produced. When I added “default shell” to /etc/kdump.conf, did a “service dump restart” to re-write the init.rd file, and re-crashed with “echo c > /proc/sysrq-trigger”, the console told me it couldn’t mount /dev/vg00/lv_kdump to mnt.
After some digging the solution to this was to amend the crashkernel=auto option in /etc/grub.conf to crashkernel=768M, reboot (for the new kernel option to take effect), and then a crash dump test actually created the /var/crash/127.0.0..etc directory and crash file.
