Description:{}

OSDFM are seeing OCM Agent pods stuck in Pending state in Stage ap-southeast-6 SC.

Symptoms:{}

• Pods: ocm-agent-operator-registry-* in openshift-ocm-agent-operator namespace

• NodeSelector on pod: node-role.kubernetes.io/infra

• Node(s) in ap-southeast-6 new SC did not have the required label

  node-role.kubernetes.io/infra

  initially.

• As a result, pods remain in Pending / FailedScheduling.

History / Attempts:{}

Other Stage/INT regions are working because their worker nodes have the label:

node-role.kubernetes.io/infra="" 

Last week, I tried adding:

ext-node-role.kubernetes.io/infra=true 

in OSDFM deploy.yml

• This allowed pods to schedule temporarily, likely as a workaround.

• Note: Latest Terraform seems to not allow node-role.kubernetes.io/infra="", it may require some non-empty value.https://redhat-internal.slack.com/archives/C035W96HKN3/p1763742427268339

Current cause:{}

• Pod nodeSelector strictly requires the key node-role.kubernetes.io/infra.

• ap-southeast-6 nodes were missing this key, hence pods could not schedule.

Action requested:{}

Investigate the reason and confirm  that all nodes in ap-southeast-6 that should run OCM agent pods have the correct label like the other region: 

node-role.kubernetes.io/infra="" or true

If cannot figure out why ap-southeast-6 nodes were missing this key, OSDFM still needs to ensure SC creation in stage and prod ap-southeast-6 before the 11/30/2025 new region enable deadline, then SRE can manually add this label to ensure at least one usable SC + MC exists in this region.