-
Bug
-
Resolution: Unresolved
-
Normal
-
4.20
-
None
-
Quality / Stability / Reliability
-
False
-
-
None
-
None
-
None
-
None
-
None
-
None
-
In Progress
-
Release Note Not Required
-
None
-
None
-
None
-
None
-
None
Description of problem:
The OpenShift installer's Nutanix platform code in pkg/asset/machines/nutanix/machinesets.go has a critical bug when generating MachineSets for compute pools with autoscaling enabled. When pool.Replicas is nil (indicating autoscaling) and failure domains are configured, the MachineSets generation logic fails to create any MachineSets, completely breaking the cluster's ability to autoscale.
Version-Release number of selected component (if applicable):
How reproducible:
Always reproducible when both conditions are met: 1. Compute pool configured with autoscaling (replicas field omitted from install-config.yaml). 2. Nutanix platform has failure domains configured in install-config.yaml
Steps to Reproduce:
1. Create install-config.yaml with Nutanix platform including failure domains 2. Configure compute pool with autoscaling (omit replicas field) 3. Run installer: openshift-install create cluster 4. Check generated MachineSets in the clusterCheck generated MachineSets in the cluster
Actual results:
- No worker MachineSets are generated in openshift-machine-api namespace - ClusterAutoscaler cannot scale the cluster (no MachineSets to target)
Expected results:
- One MachineSet should be generated per failure domain - Each MachineSet should start with 0 replicas - ClusterAutoscaler should be able to scale these MachineSets based on pod scheduling demands
Additional info:
Root Cause: In https://github.com/openshift/installer/blob/97030df02861425054b980db72d31d36de1fcb20/pkg/asset/machines/nutanix/machinesets.go#L27-L30