Bad Pods

kubectl get pods --all-namespaces -o wide | egrep -v ‘Running|Completed|Error|BackOff|OOMKilled’

Autoscaler State ConfigMap

kubectl -n kube-system describe configmap/cluster-autoscaler-status

Autoscaler logs (all pods)

kubectl logtail --namespace=kube-system kube-cluster-autoscaler --since=10m

Deleting a pod that’s stuck in “Terminating” state for more than 5 minutes

kubectl delete pod <pod-name> --force --grace-period=0

Cluster spot instances per node

kubectl get nodes -L lifecycle-status -L 'failure-domain.beta.kubernetes.io/zone' -L beta.kubernetes.io/instance-type -L aws.amazon.com/spot