Full or Near-Full OSDs Trigger Performance Collapse

critical

Resource ContentionUpdated Jan 7, 2026

When OSDs reach nearfull (default 85%) or full (default 95%) thresholds, Ceph begins throttling operations and can trigger rebalancing, severely degrading performance. Full OSDs prevent all writes and can cause cluster-wide unavailability.

Sources

How to Troubleshoot Ceph Performance Issues - OneUptimeoneuptime.com

Troubleshooting Guide Red Hat Ceph Storage 3 | Red Hat Customer Portalaccess.redhat.com

Red Hat Ceph Storage 7 Troubleshooting Guidedocs.redhat.com

Technologies:

CephThe root cause of this issue originates in Ceph

How to detect:

Monitor `ceph health detail` for 'nearfull osds' or 'full osds' warnings. Track ceph_osd_pct_used and ceph_count_full_osds/ceph_count_near_full_osds metrics. Alert when any OSD exceeds 85% utilization.

Recommended action:

Immediately delete unnecessary data or add capacity. Temporarily increase nearfull/full ratios only as emergency measure. Rebalance data across OSDs if utilization is uneven. Monitor ceph_osd_df to identify most full OSDs and their variance.