1. Proposed title of this feature request
Forcefully remove unhealthy pods controlled by PDB during Machine Update
2. What is the nature and description of the request?
During OpenShift 4 Updates it's possible that the Node Update get blocked due to Pod Disruption Budget constrains. This is fine and expected. But in case the pods part of the PDB are anyway unhealthy it adds additional effort to the OpenShift - Cluster Operator as manual intervention is required. Having functionality to remove unhealthy pods forcefully from OpenShift - Node(s) during update (no matter if they belong to a PDB or not) would help to further improve automation and avoid manual tasks and clean-up.
The key is that it should only be done for unhealthy pod. Here the impact even in a PDB would not be dramatic as things are anyways not running in the unhealthy pod
3. Why does the customer need this? (List the business requirements here)
Especially in non-production environments, it's very much possible that pods are sometimes unhealthy for a rather long time period. If updates are run frequently it adds additional effort to the Cluster Operator has OpenShift - Node require manual intervention when unhealthy pods in a PDB can't be drained. With such functionality we would improve the experience and avoid manual activity during OpenShift - Updates.
4. List any affected packages or components.
Machine Config Operator and Machine Config Daemon
- blocks
-
OCPBUGS-2353 [vsphere] Fail to drain master node if updating vsphere platform parameters with invalid credentials from console dashboard
-
- Closed
-
- is related to
-
WRKLDS-1490 Enforce PDB unhealthyEvictionPolicy in OpenShift
-
- New
-
- relates to
-
OCPBUGS-23796 not possible to drain a master node after multiple master nodes experience network disruption
-
- Closed
-