- Home
- All questions
- Question 578
CKA study material · question 578 of 1000
During a kubeadm upgrade that also restarts etcd, in-flight API requests stall. Which mitigation is documented?
Show the answer
Answer: B. Gracefully stop kube-apiserver a few seconds before running the upgrade
Draining does not stop the API server static Pod, so it keeps accepting work while etcd restarts. Sending SIGTERM to kube-apiserver first lets in-flight requests finish and shortens the disruption.
Source: Upgrading kubeadm clusters (Kubernetes) — Considerations when upgrading etcd