When we apply a new MC when a pool has not already applied a previous MC, and we prune the rendered MCs, the pool is degraded with this error:

  - lastTransitionTime: "2024-04-10T11:29:17Z"
    message: 'Node sregidor-voc1-r79pg-worker-a-jtg6r is reporting: "missing MachineConfig
      rendered-worker-9a036927fe6e36f4285a87683fa9d66b\nmachineconfig.machineconfiguration.openshift.io
      \"rendered-worker-9a036927fe6e36f4285a87683fa9d66b\" not found"'
    reason: 1 nodes are reporting degraded status on sync
    status: "True"
    type: NodeDegraded

pre-merge testing: https://github.com/openshift/oc/pull/1723
    

Always
    

    1. Create a MC
    2. Once the MC has been applied to the first node (but it is not applied to the second node) create a new MC
    3. Execute the command: oc adm prune  renderedmachineconfigs --confirm  


    

What happens is that the rendered MC in step 2 is created but it is not used by any pool (because we are still waiting for the second node to be configured), hence it is pruned. Once the second node configuration is finished, MCO tries to use the rendered MC that was created in step 2, but it has been pruned in step 3 and the pool becomes degraded

    

No pool should be degraded when we execute the "oc adm prune" command.
    

I described here the steps that I used to reproduce the issue, but it is probably easier to reproduce it with more than 2 nodes, because we don't have to pay much attention to the timing when pruning.

In order to recover the pool we need to:
1. Remove the last MC
2. Edit the desiredConfig in the degraded node
3. Edit the .spec.configuration.name value in the degraded pool