Using Pod scheduling gates

How does the operator handle Kubernetes Pod schedulingGates?

The operator supports schedulingGates under serverPod for WebLogic Server Pods. You can configure the field at the domain, Administration Server, cluster, or Managed Server scope. For a Cluster resource, configure it under spec.serverPod.schedulingGates.

When schedulingGates is configured, the operator includes it only when creating a new Pod. Kubernetes then keeps that Pod in the SchedulingGated state until the gate is removed from the Pod. The operator continues to honor normal startup ordering and concurrency limits, including starting the Administration Server before Managed Servers.

Changing or removing schedulingGates in the Domain or Cluster resource does not update existing Pods and does not cause the operator to replace them. This matches Kubernetes immutability rules for scheduling gates. To release an existing scheduling-gated Pod, remove the gate from the Pod itself, for example:

$ kubectl patch pod <pod-name> -n <namespace> --type=json \
    -p='[{"op":"remove","path":"/spec/schedulingGates"}]'

A scheduling-gated Pod may remain Pending while the gate exists. The operator does not report the Pod as failed for exceeding maxPendingWaitTimeSeconds while it is still scheduling-gated.