Update Percona Operator for PostgreSQL¶

Percona Operator for PostgreSQL allows upgrades to newer versions. This includes upgrades of the Operator itself, and upgrades of the Percona Distribution for PostgreSQL.

Upgrading the Operator¶

Only the incremental update to a nearest minor version of the Operator 1.x is supported. To update to a newer version, which differs from the current version by more than one, make several incremental updates sequentially. See documentation archive for documentation on previous versions of the Operator.

You can check the Operator images to find out the current Operator version with the following command (in case it is deployed in the pgo namespace):

$ kubectl get deployment postgres-operator -o yaml | grep percona-postgresql-operator

Expected output

image: percona/percona-postgresql-operator:1.4.0-pgo-apiserver
image: percona/percona-postgresql-operator:1.4.0-postgres-operator
image: percona/percona-postgresql-operator:1.4.0-pgo-scheduler
image: percona/percona-postgresql-operator:1.4.0-pgo-event

Note

The above command and other commands in this section follow the assumption that the context with the Operator namespace (pgo by default) was set. You can set context as follows:

$ kubectl config set-context $(kubectl config current-context) --namespace=pgo

Alternatively, you can specify the proper namespace explicitly: for example, by adding the -n pgo option to kubectl in all commands.

The following steps will update the Operator to a newer version:

Check that the Operator deployment job is not still present in your cluster:
```
$ kubectl get job/pgo-deploy -n pgo
```
Expected output
```
NAME         COMPLETIONS   DURATION   AGE
pgo-deploy   1/1           81s        5m53s
```
If the job is not present, you will get a message that it is not found. Otherwise you should delete this job before upgrading the Operator:
```
$ kubectl delete  job/pgo-deploy -n pgo
```

Upgrading the Operator is similar to deploying a new Operator version, but you should change the DEPLOY_ACTION option in the deploy/operator.yaml file before applying it from install to update:

...
  containers:
    - name: pgo-deploy
      image: percona/percona-postgresql-operator:1.4.0-pgo-deployer
      imagePullPolicy: Always
      env:
        - name: DEPLOY_ACTION
          value: update
...

You can automate this with the yq tool as follows, assuming that you are upgrading to the Operator version 1.5.1:

$ curl -s https://raw.githubusercontent.com/percona/percona-postgresql-operator/v1.5.1/deploy/operator.yaml | yq w --doc 4 - "spec.template.spec.containers[0].env[0].value" "update" | kubectl apply -f -
$ kubectl wait --for=condition=Complete job/pgo-deploy --timeout=90s

Note

The example above (and other examples in this document) uses the yq version 3.4.0. Note that the syntax for the yq command may be slightly different in other versions.

Applying the modified operator.yaml will produce the command output as follows:

serviceaccount/pgo-deployer-sa unchanged
clusterrole.rbac.authorization.k8s.io/pgo-deployer-cr unchanged
configmap/pgo-deployer-cm configured
clusterrolebinding.rbac.authorization.k8s.io/pgo-deployer-crb unchanged
job.batch/pgo-deploy created

The pgo-deploy Kubernetes Job created to carry on the Operator deployment process can take a minute or more to be completed. You can track it with the following command:
```
$ kubectl get job/pgo-deploy
```
Expected output
```
NAME         COMPLETIONS   DURATION   AGE
pgo-deploy   1/1           81s        5m53s
```
When it reaches the COMPLETIONS count of 1/1, you can safely delete the job as follows:
```
$ kubectl delete  job/pgo-deploy
```
Note

Deleting the pgo-deploy job will be needed before the next upgrade of the Operator.

Upgrading Percona Distribution for PostgreSQL¶

Automatic upgrade¶

Starting from version 1.1.0, the Operator does fully automatic upgrades to the newer versions of Percona PostgreSQL Cluster within the method named Smart Updates.

The Operator will carry on upgrades according to the following algorithm. It will query a special Version Service server at scheduled times to obtain fresh information about version numbers and valid image paths needed for the upgrade. If the current version should be upgraded, the Operator updates the CR to reflect the new image paths and carries on sequential Pods deletion in a safe order, allowing the cluster Pods to be re-deployed with the new image.

Note

Version Service is in technical preview status and is disabled by default for the Operator version 1.1.0. Disabling Version Service makes Smart Updates rely on the image keys in the Operator’s Custom Resource.

The upgrade details are set in the upgradeOptions section of the deploy/cr.yaml configuration file. Make the following edits to configure updates:

Set the apply option to one of the following values:
- recommended - automatic upgrades will choose the most recent version of software flagged as recommended (for clusters created from scratch, the Percona Distribution for PostgreSQL 14 version will be selected instead of the Percona Distribution for PostgreSQL 13 or 12 version regardless of the image path; for already existing clusters, 14 vs. 13 or 12 branch choice will be preserved),
- 14-recommended, 13-recommended, 12-recommended - same as above, but preserves specific major Percona Distribution for PostgreSQL version for newly provisioned clusters (for example, 14 will not be automatically used instead of 13),
- latest - automatic upgrades will choose the most recent version of the software available,
- 14-latest, 13-latest, 12-latest - same as above, but preserves specific major Percona Distribution for PostgreSQL version for newly provisioned clusters (for example, 14 will not be automatically used instead of 13),
- version number - specify the desired version explicitly,
- never or disabled - disable automatic upgrades
Note

When automatic upgrades are disabled by the apply option, Smart Update functionality will continue working for changes triggered by other events, such as updating a ConfigMap, rotating a password, or changing resource values.
Make sure the versionServiceEndpoint key is set to a valid Version Server URL (otherwise Smart Updates will not occur).
Percona’s Version ServiceVersion Service inside your cluster
You can use the URL of the official Percona’s Version Service (default). Set versionServiceEndpoint to https://check.percona.com.
Alternatively, you can run Version Service inside your cluster. This can be done with the kubectl command as follows:
$ kubectl run version-service --image=perconalab/version-service --env="SERVE_HTTP=true" --port 11000 --expose
Note

Version Service is never checked if automatic updates are disabled. If automatic updates are enabled, but Version Service URL can not be reached, upgrades will not occur.
Use the schedule option to specify the update checks time in CRON format.

The following example sets the midnight update checks with the official Percona’s Version Service:
```
spec:
  upgradeOptions:
    apply: recommended
    versionServiceEndpoint: https://check.percona.com
    schedule: "0 4 * * *"
...
```

Semi-automatic upgrade¶

The following command will allow you to update the Operator to current version (use the name of your cluster instead of the <cluster-name> placeholder).

$ kubectl patch perconapgcluster/<cluster-name> --type json -p '[{"op": "replace", "path": "/spec/backup/backrestRepoImage", "value": "percona/percona-postgresql-operator:1.5.1-ppg14-pgbackrest-repo"},{"op":"replace","path":"/spec/backup/image","value":"percona/percona-postgresql-operator:1.5.1-ppg14-pgbackrest"},{"op":"replace","path":"/spec/pgBadger/image","value":"percona/percona-postgresql-operator:1.5.1-ppg14-pgbadger"},{"op":"replace","path":"/spec/pgBouncer/image","value":"percona/percona-postgresql-operator:1.5.1-ppg14-pgbouncer"},{"op":"replace","path":"/spec/pgPrimary/image","value":"percona/percona-postgresql-operator:1.5.1-ppg14-postgres-ha"},{"op":"replace","path":"/spec/userLabels/pgo-version","value":"1.5.1"},{"op":"replace","path":"/metadata/labels/pgo-version","value":"1.5.1"}]'

Note

The above example is composed in assumption of using PostgreSQL 14 as a database management system. For PostgreSQL 13 you should change all occurrences of the ppg14 substring to ppg13.

This will carry on the image and the cluster version update.

Last update: 2024-05-02