Answer
Check node-level resource pressure first with kubectl describe node — a message like '0/1 nodes are available: Insufficient cpu' means the cluster node simply doesn't have enough CPU/memory headroom to schedule all the required pods. For a pod that's Evicted with a message like 'Pod ephemeral local storage usage exceeds the total limit,' that's a disk/storage-limit issue for that pod rather than a scheduling problem. Check both the main container's and the linkerd-proxy sidecar's logs (kubectl logs <pod> -c <container> -n <namespace>) for more detail, and for a pod stuck in a bad state, deleting it (kubectl delete pod <pod> -n <namespace>) lets Kubernetes recreate it cleanly, which resolves many transient stuck-pod states. Separately, ifs-db-init is a one-time Job, not a long-running pod — 'Completed' status is what you want to see there, not 'Running'.