Did you look is there error in pve system logs and looked vm cpu usage and if it stuck at a certain percentage in that case it just need full node reboot.
Or did you mean that vm stays stopped and not changing status to running
if vm host loses image and not showing "not running image" i think...