[Starlingx-discuss] How does STX detect a failed vm?

[Victor Rodriguez]

Hi team

We are testing if STX can detect a failed virtual machine. The way we
are making the VM fail is by :

$ echo 1 > /proc/sys/kernel/sysrq
$ echo c > /proc/sysrq-trigger

This will make use of the SysRq kernel mechanism to literally crush
the system [0]. After we execute this command by ssh the connection
lost and when we check the log of the VM with:

openstack console log show <VM>

We can see the kernel panic. However, when we check nova and horizon
there is no log that mentions the VM has failed; actually, the horizon
dashboard shows the VM as running. The qemu process is still running
but the VM hangs.

Do you know a way to detect this kind of failure in a VM?
What way do you recommend to make a VM fail?
killing the process of the VM running or turning it off is not the
kind of test we are looking for. Also, this could be a valid scenario
for users/costumers, where many VMs had a kernel panic but horizon
does not show anything unusual.

Thanks for all the help

Regards

Victor R

[0] http://ngelinux.com/what-is-proc-sysrq-trigger-in-linux-and-how-to-use-sysrq-kernel-feature/