Postgres major version upgrades on Hybrid Manager—Advanced High Availability or Distributed High Availability

Upgrade considerations

• Upgrade skips: Be aware that direct major version skips are not supported. All upgrades must proceed sequentially, one major version at a time.

  Correct: 16 > 17 > 18

  Incorrect: 16 > 18

• Image tag consistency: Maintain consistency in image tags. For example, upgrade from <old-image-version>-<distribution>_full to <new-image-version>-<distribution>_full. Failure to do so may result in an altered set of packaged extensions.

• Physical join: While AHA and DHA clusters use logical replication for day-to-day data synchronization, the process of a major version upgrade involves a physical join (physical replication) to facilitate the version change between data groups. This allows data groups to be upgraded in place, where they temporarily run different major versions. However, you cannot add a new data group with a different major version.

Prerequisites

The prerequisites to get started are to have a cluster created with HM up and in a healthy state.

To ensure the cluster is in a healthy state, from the cluster's main page, select the Monitoring tab. Now check the cluster and database metrics and make sure everything is operating and all metrics are within the defined acceptable parameters.

Upgrade the Postgres major version of an existing cluster created with HM

For major upgrades of Advanced High Availability or Distributed High Availability clusters (those with multiple data groups), you must upgrade all data and witness groups sequentially. While the order of groups isn't critical, ensure each groups' upgrade completes entirely before starting the next.

1. From the cluster's main page, next to the first data group you want to upgrade, select the three-dot menu, Upgrade Data Group and then choose Major Upgrade.

   The Upgrade Cluster — Major dialogue window opens.

2. Select your preferred new major version image.

3. Review the upgrade path and confirm by selecting the Continue button.

4. The dialogue window now shows the specific package changes to be expected with the upgrade. Review these and select Upgrade Cluster if everything looks as correct.

5. The upgrade is applied automatically. You can then use the PGD CLI to ensure the data group becomes healthy. The upgrade is successfull when all checks have the Status Ok.

6. After the first data group is successfully upgraded and healthy, proceed to upgrade the remaining data and witness groups one by one until the entire cluster is on the new version.

Rollback considerations

There is no automatic rollback available with this approach and performing a downgrade is not recommended. To undo the minor upgrade, restore from the latest backup in a new cluster running the preferred minor version and switching application workloads over to the new cluster. You can then delete the PG minor upgraded cluster or troubleshoot it further.