Placement Group States Reveal Data Distribution Problems
warningPGs not in 'active+clean' state (stale, degraded, inconsistent, down) indicate data availability or consistency issues. These states directly impact performance and data safety, with different states requiring different remediation approaches.
Run `ceph pg stat` to check overall PG health - healthy clusters show mostly active+clean PGs. Use `ceph pg ls | grep -v 'active+clean'` to list problematic PGs. Monitor ceph_pgstate_active_clean, ceph_count_pgs, and check for specific states like 'stale', 'inconsistent', 'down', 'unfound'.
For stale PGs: check if OSDs are down/unreachable. For inconsistent PGs: run scrub to identify discrepancies, then repair with `ceph pg repair`. For degraded PGs: verify recovery is progressing and not blocked. For down PGs: investigate OSD failures and restore quorum. Use `ceph pg X query` for detailed PG state analysis.