The disk cleanup example in the Ceph Storage Cluster with Rook documentation does not fully remove Ceph Bluestore metadata. Bluestore stores data in multiple locations on the disk, so remnants can remain even after running the documented cleanup container, which may lead to issues such as pools reporting 100% unknown PGs when the disk is reused. The documentation should note this limitation and recommend fully wiping the disk (for example with sgdisk --zap-all ${DEVICE}) when resetting OSD devices.