Comment 6 for bug 1698154

Ryan Beisner (1chb1n) wrote :

We'll be designing and implementing an action-driven approach to handle cases when a formatted device is presented for reformat. The charm will require operator decision and input in those cases.

This brings a long-standing issue with fsid re-use back to the forefront. We will need to evaluate the usefulness of explicit fsid charm config in the current world where the charms now handle that automatically. The issue with fsid re-use is that a formatted block device may contain ceph artifacts from a prior deployment, but the fsid was statically-configured, the fsid matches the current deployment, and ceph tries (unsuccessfully) to re-use it without formatting it.