List Kubernetes pods that are not Running and Ready

kubectl get pods -A on a busy cluster prints hundreds of healthy lines. This script shows only the problems: pods that are not in the Running phase, and running pods where at least one container is not ready. For each one it adds the reason Kubernetes reports, such as CrashLoopBackOff, ImagePullBackOff or OOMKilled. It only reads from the cluster.

The script

#!/usr/bin/env bash
set -euo pipefail

usage() {
  echo "Usage: $(basename "$0") [-n NAMESPACE] [-l SELECTOR] [-c]" >&2
  echo "  -n  only this namespace (default: all namespaces)" >&2
  echo "  -l  label selector, for example app=web" >&2
  echo "  -c  also list Completed pods" >&2
  exit 2
}

SCOPE=(--all-namespaces)
SELECTOR=()
SHOW_COMPLETED=0

while getopts ":n:l:ch" opt; do
  case $opt in
    n) SCOPE=(--namespace "$OPTARG") ;;
    l) SELECTOR=(--selector "$OPTARG") ;;
    c) SHOW_COMPLETED=1 ;;
    *) usage ;;
  esac
done

command -v kubectl >/dev/null || { echo "kubectl not found" >&2; exit 1; }

kubectl get pods "${SCOPE[@]}" ${SELECTOR[@]+"${SELECTOR[@]}"} \
  -o custom-columns='NAMESPACE:.metadata.namespace,NAME:.metadata.name,PHASE:.status.phase,READY:.status.containerStatuses[*].ready,WAITING:.status.containerStatuses[*].state.waiting.reason,LAST:.status.containerStatuses[*].lastState.terminated.reason,RESTARTS:.status.containerStatuses[*].restartCount,NODE:.spec.nodeName' \
  --no-headers |
awk -v show_completed="$SHOW_COMPLETED" '
  BEGIN { found = 0 }
  {
    phase = $3; ready = $4
    if (phase == "Succeeded" && show_completed == 0) next
    if (phase == "Running" && ready !~ /false/) next
    if (!found) {
      printf "%-20s %-45s %-10s %-12s %-28s %-14s %-9s %s\n", "NAMESPACE", "NAME", "PHASE", "READY", "WAITING", "LAST", "RESTARTS", "NODE"
      found = 1
    }
    printf "%-20s %-45s %-10s %-12s %-28s %-14s %-9s %s\n", $1, $2, $3, $4, $5, $6, $7, $8
  }
  END {
    if (!found) { print "All matching pods are Running and Ready." > "/dev/stderr"; exit 0 }
    exit 1
  }'

Usage

./k8s-not-running.sh
./k8s-not-running.sh -n shop
./k8s-not-running.sh -l app=web -c
./k8s-not-running.sh || echo "problems found"

The exit code is 0 when everything is healthy and 1 when at least one pod is listed, so the script can be used as a simple check in cron or CI.

Reading the columns

• PHASE: the Pod phase: Pending (not scheduled yet or still pulling images), Running, Succeeded, Failed or Unknown (node not reporting).
• READY: one true/false per container. A Running pod with false is failing its readiness probe or restarting.
• WAITING: why a container is not running right now, for example CrashLoopBackOff, ErrImagePull, ImagePullBackOff or CreateContainerConfigError (often a missing ConfigMap or Secret key).
• LAST: how the previous instance ended, for example OOMKilled or Error.
• NODE: <none> means the scheduler has not placed the pod; look at its events for a FailedScheduling message.

Limitations

• Columns that list several containers are comma-separated; for multi-container pods, check details with kubectl describe pod.
• Init containers are not shown in the READY column. A pod stuck in initialization is still listed, because its phase stays Pending.
• Pods that are being deleted can still show as Running for their grace period.

Official documentation

• Pod Lifecycle
• kubectl Quick Reference