Bug 2111364

In certain scenarios, the OSDs were slow to process RBD requests. This prevented the rbd_support module's RBD client from being able to gracefully handover a RBD exclusive lock to another RBD client. When this condition persisted, the other RBD client forcefully acquired the lock by blocklisting the module's RADOS client. Consequently, the rbd_support module stopped working, e.g., stopped scheduling mirror snapshots. The rbd_support module had to be recovered by manually restarting ceph-mgr which disrupted the operations of other mgr modules that were reloaded. On the module's RADOS client being blocklisted, instead of restarting the ceph-mgr, the rbd_support module automatically recovers. The recovery process involves shutting down the module's handlers, creating a new RADOS client for the module, and restarting the handlers. After recovery, the rbd_support module starts to serve request as before.