Skip to content

What happens when a pod is deleted

What happens behind the scene when a pod gets deleted?

When we do: 

$ kubectl delete pod nginx

pod "nginx" deleted

What actually happens behind the scenes?

Before continuing, let's remind ourselves of the linux signals that are available to us from the container perspective:

SIGTERM: Requests a graceful shutdown, allowing the program to finish tasks and clean up. In Kubernetes, pods get time to exit cleanly.

SIGKILL: Forces an immediate stop, with no cleanup. If a pod doesn't shut down in time after SIGTERM, Kubernetes sends SIGKILL to terminate it immediately.

  1. kubectl delete pod: Triggers the API Server to update ETCD with deletionTimestamp and deletionGracePeriodSeconds, marking the pod as Terminating.
  2. API Server → Kubelet: Notifies the Kubelet of the pod’s termination.
  3. Endpoint Controller: Removes the pod from active service endpoints, stopping any traffic from reaching the pod.
  4. PreStop Hook (if configured): Before sending SIGTERM, the Kubelet runs the PreStop Hook. This allows the pod to perform custom tasks (e.g., closing connections) during shutdown.
  5. SIGTERM: Kubelet sends SIGTERM, initiating a graceful shutdown. The pod is given the deletionGracePeriodSeconds (default 30s) to cleanly exit.
  6. Graceful Shutdown: During the grace period, the pod handles any ongoing tasks, such as completing requests or saving data, before it fully stops.
  7. SIGKILL: If the pod doesn't terminate within the grace period, SIGKILL is sent, forcing immediate shutdown.
  8. Pod Deleted: The API Server updates ETCD, marking the pod as deleted. Components like Kube-Proxy, Ingress, and others remove all references to the pod.