I have a FreeBSD 12.3 guest running a poller node and when it gets installed everything runs just fine. We can stop and start the guest too, no problem. The guest uses VirtIO SCSI and uses an ceph RBD image of 120GB. The FreeBSD qemu-guest-agent is installed.
If for some reason the VM is migrated to another node, the guest looses it's ability to use the disk somehow. Eventually all processes have a status "D".
I stopped the machine and made backup and then recreated the guest from the backup. I tried that because it creates a new disk image, but it makes no difference. Clearly something in either the VM Guest config or FreeBSD itself gets affected by the migration. What could it be?
How does the migration happen and what in the process can do this to the machine?
If for some reason the VM is migrated to another node, the guest looses it's ability to use the disk somehow. Eventually all processes have a status "D".
I stopped the machine and made backup and then recreated the guest from the backup. I tried that because it creates a new disk image, but it makes no difference. Clearly something in either the VM Guest config or FreeBSD itself gets affected by the migration. What could it be?
How does the migration happen and what in the process can do this to the machine?