🔹 Common PG States in Ceph

1. active + clean
   ✅ Best possible state.
   All replicas of the data are available and fully synchronized across OSDs.

Cause: The system is healthy, all OSDs are online, and replication or erasure coding is working correctly.

2. active + degraded
   🔸 Data is available, but some replicas are still missing or not yet rebuilt.

Possible causes:

• One or more OSDs are down or failed.

• Data is still being recovered on other OSDs.
  
  

3. active + recovering
   🔄 The PG is restoring data from failed OSDs to healthy ones.

Cause:

• After an OSD comes back online or a new disk is added, Ceph starts rebuilding replicas.

• High load or temporary failures.
  
  

4. active + remapped
   📦 The PG has been moved to another location due to changes in the CRUSH map or OSD reallocation.

Possible causes:

• Changes in CRUSH rules or OSD weight.

• Adding/removing OSDs.
  
  

5. active + backfilling
   🔄 Similar to recovering, but filling PGs onto a newly added or freed OSD.

Cause: A new OSD was added or CRUSH weights changed, triggering data movement.

6. active + undersized
   ⚠️ The number of available replicas is less than the required count.

Possible causes:

• Some OSDs are down or out.

• Insufficient capacity to host all replicas.
  
  

7. stale
   ⏳ The state is unknown because OSDs holding the PG haven’t reported for a long time.

Possible causes:

• Network issues between MONs and OSDs.

• OSD daemon is completely down.
  
  

8. peering
   🔍 The process of synchronizing replicas and selecting the authoritative copy.

Possible causes:

• OSD restarts.

• CRUSH changes or data movement.
  
  

9. incomplete
   ❌ Ceph could not find a complete or reliable replica.
   Data may be lost.

Possible causes:

• Too many OSDs are unavailable.

• Required quorum for authoritative copy was not reached.
  
  

10. down
    🚫 The PG is fully inactive and no replicas are available.

Possible causes:

• All OSDs containing the PG are down.

• Disk corruption or severe hardware failure.