Red Hat build of MicroShift

Pod networking cannot be configured

If the Multus CNI plugin cannot apply networking annotations to a pod, the pod does not start. Pods can also fail to start if any of the additional network CNIs fail.

Example error

Warning  NoNetworkFound     0s     multus    cannot find a network-attachment-definitio (asdasd) in namespace (default): network-attachment-definitions.k8s.cni.cncf.io "bad-ref-doesnt-exist" not found

In this case, you can take the following steps to trouble CNI failures:

Verify the values in both the NetworkAttachmentDefinitions and the annotations.
Remove the annotation to verify whether the pod is created successfully with just the default network. If not, this might indicate a networking problem other than the Multus configuration.

If you are a device administrator, you can inspect the crio.service or microshift.service logs, paying special attention to those that are generated by the kubelet.

For example, the following error from the kubelet shows that the primary CNI is not running. This situation can be caused by pods not starting or because of a CRI-O misconfiguration such as an incorrect cni_default_network setting.

Example kubelet-generated error

Feb 06 13:47:31 dev microshift[1494]: kubelet E0206 13:47:31.163290    1494 pod_workers.go:1298] "Error syncing pod, skipping" err="network is not ready: container runtime network not ready: NetworkReady=false reason:NetworkPluginNotReady message:Network plugin returns error: No CNI configuration file in /etc/cni/net.d/. Has your network provider started?" pod="default/samplepod" podUID="fe0f7f7a-8c47-4488-952b-8abc0d8e2602"