#Setting Ceph OSD Full Thresholds

#Check the current cluster state

If the rook-ceph-tools Pod is not running, start it first:

kubectl -n rook-ceph scale deploy rook-ceph-tools --replicas=1

Check overall cluster health:

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- ceph -s

Check the current threshold values:

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- \
  ceph osd dump | egrep 'nearfull_ratio|backfillfull_ratio|full_ratio'

output example:

full_ratio 0.95
backfillfull_ratio 0.9
nearfull_ratio 0.85

#Seting the thresholds via ceph CLI

Use Ceph commands to change the effective cluster values directly.

For example, to raise the thresholds slightly:

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- \
  ceph osd set-nearfull-ratio 0.88

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- \
  ceph osd set-backfillfull-ratio 0.92

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- \
  ceph osd set-full-ratio 0.96

Use the smallest increase that restores cluster progress. After changing the values, continue with capacity expansion or data cleanup as soon as possible.

If writes are blocked and OSDs remain stuck, pending, or cannot come back up, stop application I/O first, then raise only the full threshold by a small amount, wait for the rebalance to finish, and restore the threshold after the cluster returns to a stable state.

#Setting the thresholds by updating the CephCluster CR

You can set Ceph OSD full thresholds by updating the CephCluster CR, Use this procedure if you want to override the default settings.

kubectl patch cephcluster ceph-cluster -n rook-ceph --type merge -p \
'{"spec":{"storage":{"nearFullRatio":0.88,"backfillFullRatio":0.92,"fullRatio":0.96}}}'

If you only need to change one threshold, patch only that field. For example:

kubectl patch cephcluster ceph-cluster -n rook-ceph --type merge -p \
'{"spec":{"storage":{"fullRatio":0.96}}}'

#Verify the applied values

Verify the persisted settings in Kubernetes:

kubectl get cephcluster ceph-cluster -n rook-ceph -o yaml

Verify the effective runtime values in Ceph:

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- \
  ceph osd dump | egrep 'nearfull_ratio|backfillfull_ratio|full_ratio'

Recheck cluster health and rebalance status:

kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- ceph -s

When these fields are set in CephCluster, that resource becomes the declarative source for the threshold values. If ACP or Rook reconciles the cluster configuration later, the values in CephCluster should be treated as the intended baseline.

#Restore or rebaseline the thresholds

After you add capacity, remove data, or complete rebalancing, decide whether to keep the new thresholds. If the higher values were used only as an emergency workaround, patch the CephCluster resource back to your standard baseline and confirm that the runtime values also return to the expected state.