-
Story
-
Resolution: Done
-
Normal
-
None
-
None
-
BU Product Work
-
False
-
None
-
False
-
OCPSTRAT-291 - OpenShift AutoScaler (Phase 2)
-
If Release Note Needed, Set a Value
-
Set a Value
-
CLOUD Sprint 229, CLOUD Sprint 230
User Story
As an OpenShift user, I would like the cluster autoscaler to properly balance similar node groups regardless of the platform I am running on. Having the cluster autoscaler operator automatically apply the proper ignore labels flags would allow me to do this.
Background
As the upstream community would like to not carry any ignore labels inherently in the autoscaler, we should make our autoscaler operator deploy the autoscaler with the proper ignore labels specified through command line flags.
This will require us to add some logic to the CAO to detect the platform and then apply the proper label flags.
Steps
- Update CAO to carry logic for detecting the platform and then applying the appropriate labels. This can be inspired by the current nodegroupset processor that exists in the upstream as released in 1.25 and expressed in this closed PR .
Stakeholders
- openshift engineering
Definition of Done
- Users can enable the balance similar nodes options on any platform and continue to use the autoscaler as expected, with nodes balancing properly.
- Docs
- This should not require a docs update, but adding something to the developer docs for the CAO would be nice.
- Testing
- We currently have tests for this feature, they should continue to work as expected.
- links to