Force-remove a stuck pod sandbox

Clean up orphaned NotReady pod sandboxes the kubelet failed to GC; node-level recovery

crictl rmp -f $(crictl pods -q --state NotReady)
tool
crictl3 commands
tags
crictlk8stroubleshootingdebug

More crictl commands

  • Prune images on a node
    crictl rmi --prune

    Reclaim disk by deleting images not referenced by any container; safe node disk-pressure fix

  • Inspect container CRI metadata
    crictl inspect <id> | jq '.info'

    Dump runtime info, cgroup path, mounts and pid for a CRI container kubectl can't show

all 3 crictl commands