Upgrading the Postgres Operator Release

This topic provides steps to upgrade the Postgres operator currently deployed in your Kubernetes cluster to a new Postgres operator version.

Upgrading from VMWare Tanzu Postgres 0.4.0 or Later

Follow these steps to upgrade the Postgres operator version 0.4.0 or later to a newer Tanzu SQL with Postgres for Kubernetes release. Upgrading the Postgres operator does not upgrade existing Postgres instances. See Upgrading Postgres Instances for steps to upgrade instances.

  1. Follow the steps in Installing Tanzu SQL with Postgres for Kubernetes to load the new Postgres operator and Postgres instances images into your container registry.

  2. Upgrade the Postgres operator with the helm upgrade command.

    $ helm upgrade postgres-operator operator/

    Release "postgres-operator" has been upgraded. Happy Helming!
NAME: postgres-operator
LAST DEPLOYED: Tue May 19 14:59:20 2020
NAMESPACE: default
STATUS: deployed
REVISION: 2
TEST SUITE: None

  3. Verify the Postgres operator version.

    $ helm ls

    NAME                NAMESPACE   REVISION    UPDATED                                 STATUS      CHART                                       APP VERSION
postgres-operator   default     1           2020-06-16 13:28:05.704226 -0500 CDT    deployed    postgres-operator-v0.4.0

Warning Do not delete the Postgres operator when there are existing Postgres instances created by the operator. If the Postgres operator is deleted, you will get an error if you try to reinstall it. See Cannot Reinstall Operator after Deleting for a workaround to this problem.

Upgrading from VMware Tanzu Postgres 0.3.0 or Earlier

Tanzu SQL with Postgres 0.4.0 and later require that you install cert-manager in the Kubernetes cluster before you install the Postgres operator. To upgrade from version 0.3.0 or earlier, you must remove the Postgres instance’s custom resource (CRD), install cert-manager, and then upgrade the Postgres operator with Helm.

When you delete the Postgres instance’s CRD, your existing Postgres instances are also deleted. The instances’ PVCs are not deleted, however, so you just need to recreate the instances after upgrading the Postgres operator and the existing data will be available.

If you are upgrading the Postgres operator from Tanzu SQL with Postgres 0.3.0 or earlier, follow these steps:

  1. Find the name of the current Postgres operator CRD.

    $ kubectl get crds

    NAME                                    CREATED AT
postgres.sql.tanzu.vmware.com   2020-05-19T18:30:23Z

  2. List existing Postgres instances. You will need to recreate these after you upgrade the Postgres operator.

    $ kubectl get postgres

  3. Delete the CRD.

    $ kubectl delete crd postgres.sql.tanzu.vmware.com

  4. Create a certificate using cert-manager.

    $ kubectl apply --validate=false \
  -f https://github.com/jetstack/cert-manager/releases/download/v0.14.1/cert-manager.yaml

  5. Upgrade the Postgres operator.

    • Follow the steps in Installing Tanzu SQL with Postgres to load the new Postgres operator and Postgres instance images into your container registry.

    • Upgrade the Postgres operator with helm upgrade.

      $ helm upgrade postgres-operator operator/

      Release "postgres-operator" has been upgraded. Happy Helming!
NAME: postgres-operator
LAST DEPLOYED: Tue Jun 16 14:59:20 2020
NAMESPACE: default
STATUS: deployed
REVISION: 2
TEST SUITE: None

  6. Verify the Postgres operator version.

    $ helm ls

    NAME                NAMESPACE   REVISION    UPDATED                                 STATUS      CHART                                       APP VERSION
postgres-operator   default     1           2020-06-16 13:28:05.704226 -0500 CDT    deployed    postgres-operator-v0.4.0

  7. Recreate Postgres instances, using the same YAML manifest file used to create them initially. For example:

    $ kubectl create -f pg-instance-example.yaml

The next time you upgrade the Postgres operator, you only need to perform step 5.