How to manually remove bridged network interfaces


We recently had a database corruption, which made it impossible for us to use opennebula. We could not even shutdown VMs. So we decided to rollback the database to a backup. This left a few VMs basically stranded, as they were created after the backup date. That’s no problem, however, they are still listening to their IPs which creates a problem for us, as now two VMs may be assigned the same IP. Opennebula obviously does not know about the orphaned VMs. Can somebody tell me what is the correct way to clear the traces of all old VMs.

When I log into a host I can remove the interface using brctl tools (brctl delif br0 one-328-0) and then I down the interface in the ip utility( ip link set dev one-328-0 down) and delete it afterwards (sudo ip link delete one-328-0).

Is this the correct procedure?
Thanks :slight_smile:

That is the correct way of clearing the network. To clear the traces of the VM on the host, basically you have to clear up any resource used by the action of creating a VM.

  • Compute resources: deleting the VM on the hypervisor
  • Networking: what you just did
  • Storage: removing the disk images used by the stranded VMs. These are located at /var/lib/one/datastores/<system_datastore id>/<vm_id>