Was this page helpful?
Caution
You're viewing documentation for an unstable version of Scylla Operator. Switch to the latest stable version.
Upgrading ScyllaDB clusters¶
Upgrading your ScyllaDB cluster to a newer version is automated by Scylla Operator and performed using a rolling update strategy to maintain availability. It is as simple as updating the ScyllaDB image reference in your ScyllaDB cluster specification.
Warning
While the cluster remains operational throughout the process, applications requiring strict consistency levels (such as QUORUM)
may experience transient unavailability. This can occur if the cluster topology view has not yet fully converged across all
nodes before the next node is restarted.
We recommend scheduling upgrades during periods of low application traffic to minimize potential disruptions.
Issue tracking fix for this behavior: scylla-operator #1077.
Warning
ScyllaDB version upgrades must be performed consecutively, meaning you must not skip any major or minor version on the upgrade path. Before upgrading to the next version, ensure the entire ScyllaDB cluster has been successfully upgraded. For details, refer to the Upgrade procedure in ScyllaDB’s documentation.
Caution
Before upgrading ScyllaDB, ensure the target version is supported by the version of Scylla Operator you are using. Refer to the support matrix for information on version compatibility.
Upgrade via GitOps (kubectl)¶
To upgrade your ScyllaDB cluster using GitOps (kubectl), adjust the ScyllaDB image tag/reference to the target one in your ScyllaDB cluster specification and re-apply the manifest.
apiVersion: scylla.scylladb.com/v1
kind: ScyllaCluster
metadata:
name: scylladb
spec:
version: 2025.4.1 # Specify the target ScyllaDB image tag.
# ...
After reapplying the manifest, wait for your ScyllaCluster to roll out.
kubectl wait --for='condition=Progressing=False' scyllacluster.scylla.scylladb.com/scylladb
kubectl wait --for='condition=Degraded=False' scyllacluster.scylla.scylladb.com/scylladb
kubectl wait --for='condition=Available=True' scyllacluster.scylla.scylladb.com/scylladb
apiVersion: scylla.scylladb.com/v1alpha1
kind: ScyllaDBCluster
metadata:
name: dev-cluster
spec:
scyllaDB:
image: docker.io/scylladb/scylla:2025.4.1 # Specify the target ScyllaDB image reference.
# ...
After reapplying the manifest, wait for your ScyllaDBCluster to roll out.
kubectl --context=${CONTROL_PLANE_CONTEXT} wait --for='condition=Progressing=False' scylladbcluster.scylla.scylladb.com/dev-cluster
kubectl --context=${CONTROL_PLANE_CONTEXT} wait --for='condition=Degraded=False' scylladbcluster.scylla.scylladb.com/dev-cluster
kubectl --context=${CONTROL_PLANE_CONTEXT} wait --for='condition=Available=True' scylladbcluster.scylla.scylladb.com/dev-cluster
To verify the cluster state, execute nodetool status in each of the ScyllaDB cluster pods:
NAMESPACE=<namespace>
CLUSTER_NAME=<cluster-name>
# List all ScyllaDB pods in the cluster.
pods=$(kubectl -n "${NAMESPACE}" get pods \
-l scylla/cluster="${CLUSTER_NAME}" \
-l scylla-operator.scylladb.com/pod-type=scylladb-node \
-o name)
# Execute nodetool status in each pod.
for pod in $pods; do
kubectl -n "${NAMESPACE}" exec "${pod}" -c scylla -- nodetool status
done
All nodes should report all other nodes as UN (Up and Normal) in the output, e.g.:
Datacenter: us-east-1
===========================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns Host ID Rack
UN 10.221.135.48 3.30 KB 256 ? 5dd7f301-62d7-4ab7-986a-e7ea9d21be4d us-east-1a
UN 10.221.140.203 3.48 KB 256 ? 2f725f88-33fa-4ca7-b366-fa35e63e7c72 us-east-1b
UN 10.221.150.121 3.67 KB 256 ? 7063a262-fa3f-4f69-8a60-720f464b1483 us-east-1c
Upgrade via Helm¶
Important
Scylla Operator does not yet support Helm installation path for managed multi-datacenter ScyllaDB clusters.
To upgrade your ScyllaDB cluster using Helm, upgrade your Helm release with the target ScyllaDB image tag/reference.
helm upgrade scylla scylla/scylla --reuse-values --set=scyllaImage.tag=2025.4.1
After upgrading the release, wait for your ScyllaCluster to roll out.
kubectl wait --for='condition=Progressing=False' scyllacluster.scylla.scylladb.com/scylladb
kubectl wait --for='condition=Degraded=False' scyllacluster.scylla.scylladb.com/scylladb
kubectl wait --for='condition=Available=True' scyllacluster.scylla.scylladb.com/scylladb
To verify the cluster state, execute nodetool status in each of the ScyllaDB cluster pods:
NAMESPACE=<namespace>
CLUSTER_NAME=<cluster-name>
# List all ScyllaDB pods in the cluster.
pods=$(kubectl -n "${NAMESPACE}" get pods \
-l scylla/cluster="${CLUSTER_NAME}" \
-l scylla-operator.scylladb.com/pod-type=scylladb-node \
-o name)
# Execute nodetool status in each pod.
for pod in $pods; do
kubectl -n "${NAMESPACE}" exec "${pod}" -c scylla -- nodetool status
done
All nodes should report all other nodes as UN (Up and Normal) in the output, e.g.:
Datacenter: us-east-1
===========================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns Host ID Rack
UN 10.221.135.48 3.30 KB 256 ? 5dd7f301-62d7-4ab7-986a-e7ea9d21be4d us-east-1a
UN 10.221.140.203 3.48 KB 256 ? 2f725f88-33fa-4ca7-b366-fa35e63e7c72 us-east-1b
UN 10.221.150.121 3.67 KB 256 ? 7063a262-fa3f-4f69-8a60-720f464b1483 us-east-1c