Crane OCP e2e test run, failure summary
The downstream build test run resulted in 12 failures across 4 root causes:
crane validate offline mode flag conflict — 4 tests fail because crane validate rejects the combination of --api-resources and --kubeconfig as mutually exclusive. This affects all offline-mode validate tests.
OCP golden-file baseline drift — 2 tests fail because OCP namespaces include additional resources not present in the minikube-based golden baseline: OCP-specific RoleBindings (system:deployers, system:image-builders, system:image-pullers), builder/deployer ServiceAccounts, dockercfg Secrets, and the openshift-service-ca.crt ConfigMap. The golden files need to be regenerated against an OCP cluster.
OCP SCC UID rejection — 2 tests fail because the app-with-empty-pvc pod specifies runAsUser: 1000, which falls outside the OCP cluster's allowed UID range. The app deployer role needs to be updated to omit the fixed UID and let OCP assign one via SCC.
Stateful app deployment failures — 4 tests fail at the app setup stage before migration is attempted. MySQL and MongoDB pods never reach ready state, the cronjob PVC remains unbound on gp3-csi, and the Route validation test reports an unexpected result against the target cluster. These may point to environment or cluster configuration issues rather than crane itself.
Crane OCP e2e test run, failure summary
The downstream build test run resulted in 12 failures across 4 root causes:
crane validateoffline mode flag conflict — 4 tests fail becausecrane validaterejects the combination of--api-resourcesand--kubeconfigas mutually exclusive. This affects all offline-mode validate tests.OCP golden-file baseline drift — 2 tests fail because OCP namespaces include additional resources not present in the minikube-based golden baseline: OCP-specific RoleBindings (
system:deployers,system:image-builders,system:image-pullers), builder/deployer ServiceAccounts, dockercfg Secrets, and theopenshift-service-ca.crtConfigMap. The golden files need to be regenerated against an OCP cluster.OCP SCC UID rejection — 2 tests fail because the
app-with-empty-pvcpod specifiesrunAsUser: 1000, which falls outside the OCP cluster's allowed UID range. The app deployer role needs to be updated to omit the fixed UID and let OCP assign one via SCC.Stateful app deployment failures — 4 tests fail at the app setup stage before migration is attempted. MySQL and MongoDB pods never reach ready state, the cronjob PVC remains unbound on
gp3-csi, and the Route validation test reports an unexpected result against the target cluster. These may point to environment or cluster configuration issues rather than crane itself.