clever-processor-7873601/24/2023, 7:15 AM
, what we're trying to understand is what changes to e.g. the
resource trigger drain and cordon operations. I understand that there's a couple of components involved here that may play a role in this. AFAICT we have: 1.
2. In Rancher Cluster API with the Cluster API Provider OpenStack since our cloud is built ontop OpenStack 3. OpenStack The Terraform provider might be the component in this that has the least to say in whether a instance gets recreated or cordon+drained due to a change. So on a higher level we're more or less asking: "What changes to resources in the Terraform provider triggers what operations on the instances?". Is there a deterministic way of knowing this in beforehand, is it documented somewhere? As an example, we enabled the etcd snapshot feature (to S3) with these settings in the
in a already deployed cluster, somewhat surprising we noticed that this triggered a cordon + drain of all nodes including worker nodes. We expected this to trigger some kind of operation on the control plane nodes only. Some changes, that changes certain settings on an instance level in OpenStack might lead to a recreate VM operation, and that would be documented in OpenStack. But these more subtle changes is hard to find information on. Any insight in this would be highly appreciated!