Question

A team reports that new replicas of their Deployment stay in Pending status after they scaled it up. The cluster has three worker nodes, all showing status Ready. Running 'kubectl describe pod' on a pending pod shows the event: '0/3 nodes are available: 1 node(s) had untolerated taint {node.kubernetes.io/unschedulable: }, 2 node(s) didn't match Pod's node affinity/selector.' The pods have a nodeSelector of 'disktype=ssd'. What is the MOST likely combined cause and remediation?

Accepted Answer

One node was cordoned (marked unschedulable) and the other two nodes lack the 'disktype=ssd' label; uncordon the node or label the appropriate nodes so a schedulable node matches the selector.

Answer

The kube-scheduler is down; restart the scheduler static pod so the pending pods can be assigned to nodes.

Answer

The pods request more CPU and memory than any node can provide; lower the resource requests in the Deployment spec.

Answer

CoreDNS is failing to resolve the node names; restart the CoreDNS Deployment to restore scheduling.