We had a discussion with Rodrigo, Dan and Alex where we agreed that the path forward to fix this issue is the following:
Check if nova-compute is related to ceph-mon and configured with libvirt-image-backend=rbd [0], then [workarounds].ensure_libvirt_rbd_instance_dir_cleanup is set to True, unless instances-path[1] is set to a non-default value where the charm will go into blocked state informing the user about the unsupported configuration.
We agreed this will improve the situation for operators using Charmed OpenStack while still putting safeguards to minimize the risk of losing data.
We had a discussion with Rodrigo, Dan and Alex where we agreed that the path forward to fix this issue is the following:
Check if nova-compute is related to ceph-mon and configured with libvirt- image-backend= rbd [0], then [workarounds] .ensure_ libvirt_ rbd_instance_ dir_cleanup is set to True, unless instances-path[1] is set to a non-default value where the charm will go into blocked state informing the user about the unsupported configuration.
We agreed this will improve the situation for operators using Charmed OpenStack while still putting safeguards to minimize the risk of losing data.
[0] https:/ /charmhub. io/nova- compute/ configure# libvirt- image-backend /charmhub. io/nova- compute/ configure# instances- path
[1] https:/