Was this page helpful?
This page describes Scylla Operator upgrade procedures.
There are two generic update procedures - via Helm and via kubectl. Before upgrading, please check this page to find out
if your target version requires additional upgrade steps.
Helm doesn’t support managing CustomResourceDefinition resources (#5871, #7735)
These are only created on first install and never updated. In order to update them, users have to do it manually.
Replace <release_name>
with the name of your Helm release for Scylla Operator and replace <version>
with the version number you want to install:
Make sure Helm chart repository is up-to-date:
helm repo add scylla-operator https://storage.googleapis.com/scylla-operator-charts/stable
helm repo update
Update CRD resources. We recommend using --server-side
flag for kubectl apply
, if your version supports it.
tmpdir=$( mktemp -d ) \
&& helm pull scylla-operator/scylla-operator --version <version> --untar --untardir "${tmpdir}" \
&& find "${tmpdir}"/scylla-operator/crds/ -name '*.yaml' -printf '-f=%p ' \
| xargs kubectl apply
Update Scylla Operator
helm upgrade --version <version> <release_name> scylla-operator/scylla-operator
Replace <version>
with the version number you want to install:
Checkout source code of version you want to use:
git checkout <version>
Manifests use rolling minor version tag, you may want to pin it to specific version:
find deploy/operator -name "*.yaml" | xargs sed --follow-symlinks -i -E "s^docker.io/scylladb/scylla-operator:[0-9]+\.[0-9]+^docker.io/scylladb/scylla-operator:<version>^g"
Update Scylla Operator. We recommend using --server-side
flag for kubectl apply
, if your version supports it.
kubectl apply -f deploy/operator
v1.2.0
-> v1.3.0
¶Sidecar image is going to be upgraded automatically, so a rolling restart of your Scylla clusters is expected during the upgrade procedure.
Checkout source code of v1.3.0:
git checkout v1.3.0
Update Scylla Operator from deploy directory:
kubectl -n scylla-operator apply -f deploy/operator
Wait until Scylla Operator is up and running:
kubectl wait --for condition=established crd/scyllaclusters.scylla.scylladb.com
kubectl -n scylla-operator rollout status deployment.apps/scylla-operator
v1.1.0
-> v1.2.0
¶1.2.0 release brought a lot of changes to the Scylla Operator deployment process. To properly update Scylla Operator one must delete old objects and install updated ones.
Sidecar image is going to be upgraded automatically, so a rolling restart of your Scylla clusters is expected during the upgrade procedure.
Checkout source code of v1.2.0:
git checkout v1.2.0
Remove old scylla operator namespace - in our case it’s called scylla-operator-system
:
kubectl delete namespace scylla-operator-system --wait=true
Remove old webhooks:
kubectl delete MutatingWebhookConfiguration scylla-operator-mutating-webhook-configuration
kubectl delete ValidatingWebhookConfiguration scylla-operator-validating-webhook-configuration
Install Scylla Operator from deploy directory:
kubectl -n scylla-operator apply -f deploy/operator
Wait until Scylla Operator is up and running:
kubectl wait --for condition=established crd/scyllaclusters.scylla.scylladb.com
kubectl -n scylla-operator rollout status deployment.apps/scylla-operator
v1.0.0
-> v1.1.0
¶During this update we will change probes and image for Scylla Operator. A new version brings an automation for upgrade of sidecar image, so a rolling restart of managed Scylla clusters is expected.
Get name of StatefulSet managing Scylla Operator
kubectl --namespace scylla-operator-system get sts --selector="control-plane=controller-manager"
NAME READY AGE
scylla-operator-controller-manager 1/1 95m
Change probes and used container image by applying following patch:
spec:
template:
spec:
containers:
- name: manager
image: docker.io/scylladb/scylla-operator:1.1.0
livenessProbe:
httpGet:
path: /healthz
port: 8080
scheme: HTTP
readinessProbe:
$retainKeys:
- httpGet
httpGet:
path: /readyz
port: 8080
scheme: HTTP
To apply above patch save it to file (operator-patch.yaml
for example) and apply to Operator StatefulSet:
kubectl -n scylla-operator-system patch sts scylla-operator-controller-manager --patch "$(cat operator-patch.yaml)"
v0.3.0
-> v1.0.0
¶Note: There’s an experimental migration procedure available here.
v0.3.0
used a very common name as a CRD kind (Cluster
). In v1.0.0
this issue was solved by using less common
kind which is easier to disambiguate. (ScyllaCluster
).
This change is backward incompatible, so Scylla cluster must be turned off and recreated from scratch.
In case you need to preserve your data, refer to backup and restore guide.
Get list of existing Scylla clusters
kubectl -n scylla get cluster.scylla.scylladb.com
NAME AGE
simple-cluster 30m
Delete each one of them
kubectl -n scylla delete cluster.scylla.scylladb.com simple-cluster
Make sure you’re on v0.3.0
branch
git checkout v0.3.0
Delete existing CRD and Operator
kubectl delete -f examples/generic/operator.yaml
Checkout v1.0.0
version
git checkout v1.0.0
Install new CRD and Scylla Operator
kubectl apply -f deploy/operator.yaml
Migrate your existing Scylla Cluster definition. Change apiVersion
and kind
from:
apiVersion: scylla.scylladb.com/v1alpha1
kind: Cluster
to:
apiVersion: scylla.scylladb.com/v1
kind: ScyllaCluster
Once your cluster definition is ready, use kubectl apply
to install fresh Scylla cluster.
Was this page helpful?