In claudie we allow to add/delete loadbalancers, add/delete nodepools and along those changes the api endpoint can be changed aswell. If however an api endpoint change is to be executed and the given Task fails any subsequent changes will fail also due to the incomplete api endpoint change and the cluster will be left in an unusable state.
Ideally we would have an endless retry strategy for tasks that handle the api endpoint change. If something in the cluster is wrong it would need to be resolved manually to "unblock" the api endpoint change task so that the cluster does not end up in a malformed state.
Description
In claudie we allow to add/delete loadbalancers, add/delete nodepools and along those changes the api endpoint can be changed aswell. If however an api endpoint change is to be executed and the given Task fails any subsequent changes will fail also due to the incomplete api endpoint change and the cluster will be left in an unusable state.
Ideally we would have an endless retry strategy for tasks that handle the api endpoint change. If something in the cluster is wrong it would need to be resolved manually to "unblock" the api endpoint change task so that the cluster does not end up in a malformed state.
Exit criteria