I'm running a sometimes highly loaded Debian 7 VM with the latest 3.2.x kernel on a XenServer 6.2 host.
Sometimes this VM runs into the following state:
- Ping is okay, All Services unreachable (including SSH)
- A message showing a trace + "kernel BUG" shows up
- Further below the error message “fixing recursive fault but reboot is needed” is shown
kernel.panic=1 is already set. This has no effect in this case. The machine does not reboot.
How is is possible to let the machine reboot itself after such a fault (since this is no OOPS, right?)?
Ideally before rebooting a sync of filesystems and ro mount would be nice...
I know about SysRQ but I'm searching a solution which does work without manual intervention.