The request serving node placement needs the hosted-cluster-size label.
We had an HCP missing this label and consequently hypershift-operator is unable to assign request serving nodes. 

labels: 
    hypershift.openshift.io/hosted-cluster-size: ""

The HCP's pull secret was no longer valid and could not be used to pull release images.

Manually setting the label allowed for nodes to be choosen, but breaking the reconcile for hypershift operator ( field was no longer managed by hypershift operator )
To fix we patched the manageFields, which allowed hypershift operator to once again manage the field. 

oc patch  hostedclusters --type=json -p='[{"op": "remove", "path": "/metadata/managedFields/3"}]' 
 
So we now have a hostedcluster with this label set to an empty string even though it has two request serving nodes (m5.xlarge)


The field is managed by hypershift operator. 
- apiVersion: hypershift.openshift.io/v1beta1
    fieldsType: FieldsV1
    fieldsV1:
      f:metadata:
        f:labels:
          f:hypershift.openshift.io/hosted-cluster-size: {}
    manager: hostedclustersizing
    operation: Apply
    time: "2024-11-07T12:29:33Z"

quay.io/acm-d/rhtap-hypershift-operator@sha256:2fb13139ae58d9b3b1af62b08f6321d7422665b50249e90caf1e2370ee5e9d50

Not sure how to reproduce this.
If I delete one request serving node, the controlplane components degrade and do not land on the replacement serving node.

    1. In a request sizing management cluster, create a hosted cluster.
    2. Wait for the hosted cluster to come up.
    3. Replace the hosted cluster's pull secret with an invalid pull secret.
    

    The hosted-cluster-size label is empty string 

    The hosted-cluster-size label should have a valid value 

The worker machine set for this HCP has two workers of which both are unavailable.