When we apply a PIS to the worker pool and a custom pool exists in the cluster, the nodes that belong to the custom pool report a failure in the MCD

I0307 17:59:22.343736    2544 pinned_image_set.go:304] Reconciling pinned image set: tc-73623-worker-pinned-images: generation: 1
I0307 17:59:22.343864    2544 pinned_image_set.go:304] Reconciling pinned image set: tc-73623-worker-pinned-images: generation: 1
I0307 17:59:22.345356    2544 pinned_image_set.go:529] CRI-O config file is up to date, no reload required
E0307 17:59:22.357626    2544 upgrade_monitor.go:289] Error applying MCN status: failed to create typed patch object (/ip-10-0-23-33.us-east-2.compute.internal; machineconfiguration.openshift.io/v1alpha1, Kind=MachineConfigNode): .status.pinnedImageSets: duplicate entries for key [name="tc-73623-worker-pinned-images"]
E0307 17:59:22.357638    2544 pinned_image_set.go:607] Failed to updated machine config node: failed to create typed patch object (/ip-10-0-23-33.us-east-2.compute.internal; machineconfiguration.openshift.io/v1alpha1, Kind=MachineConfigNode): .status.pinnedImageSets: duplicate entries for key [name="tc-73623-worker-pinned-images"]


And in the MCN

  pinnedImageSets:
  - currentGeneration: 0
    desiredGeneration: 1
    lastFailedGeneration: 1
    lastFailedGenerationErrors:
    - 'failed to get image status for "quay.io/openshifttest/busybox@sha256:0415f56ccc05526f2af5a7ae8654baec97d4a614f24736e8eef41a4591f08019":
      rpc error: code = Canceled desc = context canceled'
    name: tc-73623-worker-pinned-images



Nevertheless, we can see the images pinned in the nodes

sh-5.1# crictl images --pinned
IMAGE                                                    TAG                 IMAGE ID            SIZE                PINNED
quay.io/openshifttest/alpine                             <none>              45683da4f97c2       5.87MB              true
quay.io/openshifttest/busybox                            <none>              b97242f89c8a2       1.45MB              true


    

IPI on AWS
4.19.0-0.test-2025-03-06-132141-ci-ln-2hfyl8b-latest

We use the pre-merge version corresponding to the PR that enables the use of custom pools for MCN https://issues.redhat.com/browse/MCO-1501

    

Always
    

    1. Create an infra custom pool
apiVersion: machineconfiguration.openshift.io/v1
kind: MachineConfigPool
metadata:
  name: infra
spec:
  machineConfigSelector:
    matchExpressions:
      - {key: machineconfiguration.openshift.io/role, operator: In, values: [worker,infra]}
  nodeSelector:
    matchLabels:
      node-role.kubernetes.io/infra: ""

    2. Add a node to the infra pool

oc label node $(oc get nodes -l node-role.kubernetes.io/worker -ojsonpath="{.items[0].metadata.name}") node-role.kubernetes.io/infra=

    3. Create a PIS to pin some images in all the nodes with the "worker" label

 oc create -f - << EOF
apiVersion: machineconfiguration.openshift.io/v1alpha1
kind: PinnedImageSet
metadata:
  labels:
    machineconfiguration.openshift.io/role: worker
  name: tc-73623-worker-pinned-images
spec:
  pinnedImages:
  - name: "quay.io/openshifttest/busybox@sha256:0415f56ccc05526f2af5a7ae8654baec97d4a614f24736e8eef41a4591f08019"
  - name: quay.io/openshifttest/alpine@sha256:be92b18a369e989a6e86ac840b7f23ce0052467de551b064796d67280dfa06d5
EOF


    

The images are correctly pinned in the nodes belonging to the worker pool. However, the node belonging to the infra pool reports and error in MCD and in its MCN

Images are pinned in the custom pool too. The problem seems to be the reported errors only.

    

Now that we support custom pools in MCN, we can support custom pools in PIS too.

MCO will use the pools machineconfig selector to find which PIS should be applied to each pool. Hence, the PIS that we used to reproduce this issue should be applied to the infra pool as well. And they are correctly pinned in the custom pool's node.

However, the nodes belonging to the custom pools should not report a failure due to duplicated images in the MCD and another error due to "context cancelled" in the MCN

    

Note that it seems that there is no way to apply a PIS to the worker pool but not to the custom pools, since we select the PIS the same way we select the MCs, and by design all custom pools have to use all worker pool's MCs.