Since https://issues.redhat.com/browse/OCPBUGS-36810, when the MOSC resource is deleted while the build is being executed, then all configmaps should be garbage collected. We should be able to create a new MOSC after this and the build should succeed.

When the MOSC resource is removed just before the build-rendered pod finishes, the build-rendered pod (in Terminating status) is able to create the "digest-rendered" configmap and this configmap is not garbage collected. If we create a new MOSC after this the build will fail because the new "digest-rendered" configmap will collide with the previous one. 

    

4.18.0-0.nightly-2024-09-26-093014
    

Always
    

1. Create CustomMCP
2. Apply any MOSC
3. Run this script to remove the build pod when he is about to finish

$ cat reproduce_script.sh

while true 
do
 oc logs -l machineconfiguration.openshift.io/rendered-machine-config -n openshift-machine-config-operator |grep  "buildah push"
if [ "$?" == "0" ]
then
    
    echo "Render build pod started to push the image"
    sleep 5
    echo "Remove the mosc"
    oc delete machineosconfig --all	
fi
sleep 1
done



    

The MOSB will be garbage collected, and all existing configmaps will be garbage collected. Nevertheless, the build pod will be in Terminating status for a while, and it will have time enough to create the "digest-rendered" configmap after all the garbage collection process has finished. The result is that the "digest-rendered" configmap is leaked and if we create a new MOSC the builds will fail because the new "digest-rendered" configmap will collide with the already existing one.

The result is this one

$ oc get pods,cm                                                                                                                                                                    preserve-sregidor-work: Thu Sep 26 14:59:07 2024

NAME                                                                 READY   STATUS    RESTARTS        AGE
pod/kube-rbac-proxy-crio-ip-10-0-10-89.us-east-2.compute.internal    1/1     Running   4               4h50m
pod/kube-rbac-proxy-crio-ip-10-0-41-218.us-east-2.compute.internal   1/1     Running   4               4h49m
pod/kube-rbac-proxy-crio-ip-10-0-56-75.us-east-2.compute.internal    1/1     Running   3               4h44m
pod/kube-rbac-proxy-crio-ip-10-0-8-214.us-east-2.compute.internal    1/1     Running   4               3h51m
pod/kube-rbac-proxy-crio-ip-10-0-83-194.us-east-2.compute.internal   1/1     Running   4               4h50m
pod/machine-config-controller-68df57cdf-ljrk8                        2/2     Running   0               98m
pod/machine-config-daemon-5gdj2                                      2/2     Running   2 (6m53s ago)   12m
pod/machine-config-daemon-cd7hg                                      2/2     Running   1 (9m56s ago)   12m
pod/machine-config-daemon-r97kw                                      2/2     Running   0               12m
pod/machine-config-daemon-wlbsq                                      2/2     Running   1 (9m47s ago)   12m
pod/machine-config-daemon-xpzf6                                      2/2     Running   4 (49s ago)     12m
pod/machine-config-operator-59746c6f74-wsfhq                         2/2     Running   0               91m
pod/machine-config-server-8db8k                                      1/1     Running   1               4h47m
pod/machine-config-server-fkf4r                                      1/1     Running   1               4h47m
pod/machine-config-server-vlqfv                                      1/1     Running   1               4h47m

NAME                                                               DATA   AGE
configmap/coreos-bootimages                                        4	  4h50m
configmap/digest-rendered-infra-89d689d461245f3557da375e30cc5c90   1	  12m                  <----- LEAKED
configmap/kube-rbac-proxy                                          1	  4h50m
configmap/kube-root-ca.crt                                         1	  4h51m
configmap/kubeconfig-data                                          1	  4h47m
configmap/machine-config-operator-images                           1	  4h50m
configmap/machine-config-osimageurl                                4	  4h50m
configmap/openshift-service-ca.crt                                 1	  4h51m


    

After we remove the MOSC resource all configmaps should be garbage collected and if we create a new MOSC later then the new build should not fail.