Summary

In this chapter, you have learned how to troubleshoot Ceph when all looks lost. In the event that Ceph is unable to recover PGs itself, you now understand how to manually rebuild PGs from failed OSDs. You can also rebuild the monitor's database in the event that you lose all of your monitor nodes but still have access to your OSDs. In the event of a complete cluster failure that you are unable to recover from, you also have gone through the process of recreating RBDs from the raw data remaining on your OSDs. Finally, you have configured two separate Ceph clusters and configured replication between them using RBD mirroring to provide a failover option, should you encounter a complete Ceph cluster failure.