GKE upgrade breaks postgres cluster

[vedantthapa]

There was an occurrence previously when a maintenance window was added to the GKE Autopilot Cluster. After maintenance was completed, the cloudnative-pg cluster ended up in a broken state. This might be due to the fact that cloudnative-pg operator expects to have nodeMaintenanceWindow.inProgress field set to true on the Cluster resource whenever there's a kubernetes upgrade (See docs).

We've removed the maintenance window for now so that no automated updates are triggered, however, a solution is needed to either automatically update the field in the Cluster resource before a GKE maintenance window and toggle it back to the orginial value once the upgrade is finished OR alert the user to the availability of an upgrade so they can modify the field, start the upgrade, and then toggle the field back to its initial value.

[vedantthapa]

Setting nodeMaintenanceWindow.inProgress to true might cause issues with the Cluster as per the warning mentioned on this page: https://cloudnative-pg.io/documentation/1.15/kubernetes_upgrade/#kubernetes-upgrade

[vedantthapa]

Another idea is to set the maintenance window and have a github action workflow run on a schedule before and after the maintenance window to update the relevant field in the manifest.