-
Epic
-
Resolution: Unresolved
-
Minor
-
None
-
None
-
None
-
Baremetal IPI for Power
-
False
-
None
-
False
-
Not Selected
-
NEW
-
ppc64le
-
NEW
In AI, the VIP_API/INGRESS-VIP (Cluster managed network) supports using platform.baremetal, but the OCP installation failed at the last step finalizing.
The cause of the fail is that the machine-api operator could not be initialized.
Here are the failed pods:
openshift-kube-controller-manager installer-5-master-1 0/1 Error 0 25m openshift-kube-controller-manager installer-6-master-2 0/1 Error 0 17m openshift-machine-api ironic-proxy-kgm9g 0/1 CreateContainerError 0 32m openshift-machine-api ironic-proxy-nc2lz 0/1 CreateContainerError 0 8m37s openshift-machine-api ironic-proxy-pp92t 0/1 CreateContainerError 0 32m openshift-machine-api metal3-69b945c7ff-45hqn 1/5 CreateContainerError 0 33m openshift-machine-api metal3-image-customization-7f6c8978cf-lxbj7 0/1 CreateContainerError 0 32m
the messages from failed pod ironic-proxy-nc2lz:
Normal Pulled 11m kubelet Successfully pulled image "quay.io/openshift-release-dev/ocp-v4.0-art-dev@sha256:4f84fd895186b28af912eea42aba1276dec98c814a79310c833202960cf05407" in 1.29310959s (1.293135461s including waiting) Warning Failed 11m kubelet Error: container create failed: time="2023-04-06T15:16:19Z" level=error msg="runc create failed: unable to start container process: exec: \"/bin/runironic-proxy\": stat /bin/runironic-proxy: no such file or directory"
similar errors for other failed pods.The interesting thing is some of the installation got installed in AI successfully, but these pods still are in error state.