Before You Begin

Review the Rook v1.19 release notes for any breaking changes. Key checks before upgrading:

• Confirm the current cluster is healthy: ceph status should show HEALTH_OK.
• Ensure you are running Rook v1.18 (not an older version; always upgrade one minor version at a time).
• Kubernetes cluster must meet the minimum version requirement for Rook v1.19.

kubectl -n rook-ceph exec -it <toolbox-pod> -- ceph status
kubectl -n rook-ceph get deployments rook-ceph-operator \
  -o jsonpath='{.spec.template.spec.containers[0].image}'

Option 1: Upgrade via Helm

If Rook was installed with Helm, update the chart and upgrade:

helm repo update

helm upgrade rook-ceph rook-release/rook-ceph \
  --namespace rook-ceph \
  --version 1.19.0

Check that the operator pod is replaced:

kubectl -n rook-ceph get pods -l app=rook-ceph-operator

Option 2: Upgrade via Manifests

Download the v1.19 manifests from the Rook repository:

curl -O https://raw.githubusercontent.com/rook/rook/v1.19.0/deploy/examples/crds.yaml
curl -O https://raw.githubusercontent.com/rook/rook/v1.19.0/deploy/examples/common.yaml
curl -O https://raw.githubusercontent.com/rook/rook/v1.19.0/deploy/examples/operator.yaml

Apply the CRD updates first:

kubectl apply -f crds.yaml

Apply common resources (RBAC, service accounts):

kubectl apply -f common.yaml

Update the operator deployment:

kubectl apply -f operator.yaml

Watch the operator restart:

kubectl -n rook-ceph rollout status deployment/rook-ceph-operator

Verifying the Operator Upgrade

Confirm the operator image version:

kubectl -n rook-ceph get deployment rook-ceph-operator \
  -o jsonpath='{.spec.template.spec.containers[0].image}'

Expected output should contain v1.19.0.

Upgrading the Ceph Cluster After the Operator

The operator upgrade alone does not change the running Ceph version. After the operator is running on v1.19, you can trigger a Ceph version upgrade by patching the CephCluster image:

kubectl -n rook-ceph patch cephcluster rook-ceph \
  --type merge \
  -p '{"spec":{"cephVersion":{"image":"quay.io/ceph/ceph:v19.2.0"}}}'

Monitor the rolling upgrade of OSDs, monitors, and managers:

kubectl -n rook-ceph get pods -w

Monitoring During Upgrade

Watch the cluster health throughout the process:

watch kubectl -n rook-ceph exec <toolbox-pod> -- ceph status

It is normal for health to temporarily show HEALTH_WARN during an OSD rolling restart. Wait for it to return to HEALTH_OK before proceeding.

Rollback Considerations

If you need to roll back the operator, re-apply the v1.18 manifests. The operator manages the cluster but does not change on-disk Ceph data structures, so a rollback is generally safe as long as the Ceph version itself was not upgraded.

Summary

Upgrading Rook from v1.18 to v1.19 follows a clear sequence: verify cluster health, apply updated CRDs and RBAC, then update the operator deployment. The Ceph daemons continue running during the operator upgrade. After confirming the new operator is healthy, optionally update the Ceph version by patching the CephCluster spec.