We observed that whenever a node pool is deleted in our GKE cluster, the complete network routes are also being removed automatically. Not sure if this is meant to be intentional, but it has caused disruptions in our cluster’s connectivity. Is this expected behavior? If so, how do we ensure critical routes are retained during node pool operations.
On [December 11, 2024 13:00 hrs], our gke network routes were deleted without any resource removal or configuration changes on our end. We didn’t touch the node pools around this time.
We can’t run a prod environment without understanding this behavior. Any inputs are welcome.