Error adding disk after expanding cluster

[gnuoy]

As part of sunbeam deployment the initial microceph cluster consists of one unit. More units are added and then an action on charm-microceph is run to add disks. This in turn calls:

 microceph disk add device_id

This often fails with the following error:

Error: Failed adding new disk: Failed to remove microceph_auto_osd rule: Failed to run: ceph osd crush rule rm microceph_auto_osd: exit status 16 (Error EBUSY: crush rule microceph_auto_osd (0) is in use)