Stuck in ContainerCreating
ContainerCreating means the scheduler has done its job and the node is preparing the pod. Being stuck there means one of the node-side steps is failing or slow. Events tell you which.
Diagnose in order
Section titled “Diagnose in order”kubectl describe pod <pod>and read the last events. The reason names the step.- Note the node:
kubectl get pod <pod> -o wide. If only one node is affected, the problem is that node. - For sandbox and network errors, check the node’s Ready condition and the CNI pods on that node:
kubectl get pods -n kube-system -o wide --field-selector spec.nodeName=<node>.
Usual causes
Section titled “Usual causes”| Event reason | Cause |
|---|---|
FailedCreatePodSandBox |
CNI not running or failing on the node, or no pod IPs left in the node’s range |
FailedMount, FailedAttachVolume |
A volume can’t be attached or mounted. See FailedMount |
Pulling for a long time |
A very large image or a slow registry |
| No events at all, for a long time | The kubelet on that node isn’t processing pods; check the node itself |
Fix the component the event names. If one node is the problem, cordon it (kubectl cordon <node>) and delete the stuck pod so its controller recreates it elsewhere.
Read next: How a pod gets from YAML to running.