How to fix etcd quota exceeded, event limit reached, API server rejecting requests in Kubernetes

KubernetesADVANCEDCRITICALUpdated Dec 6, 2025· 126 views

etcd database full with Kubernetes events. Fix by cleaning up old events, increasing quota, or using separate etcd for events.

Quick fix

Fixes EVENT_LIMIT_EXCEEDED

kubectl exec -n kube-system etcd-<node> -- etcdctl endpoint status
  1. Check etcd usage
  2. Clean up old events
  3. Increase etcd quota
  4. Configure event TTL
kubernetes — error
kubectl exec -n kube-system etcd-<node> -- etcdctl endpoint status
etcd quota exceeded, event limit reached, API server rejecting requests

What this error means

Kubernetes events can consume most of etcd storage, causing the cluster to reach quota and rejecting new writes.

Typical symptoms

  • Cluster becomes read-only
  • New resource creation fails
  • etcd database very large
  • API server errors about quota

Common causes

  1. Too many events generated
  2. Event cleanup/TTL not configured
  3. etcd quota too small
  4. No automatic event cleanup

How to fix "etcd quota exceeded, event limit reached, API server rejecting requests"

1Check etcd usage
bash
kubectl exec -n kube-system etcd-<node> -- etcdctl endpoint status
2Clean up old events
bash
kubectl delete events --all -A
3Increase etcd quota

Add --quota-backend-bytes flag to etcd.

4Configure event TTL

Set --event-ttl=1h on API server.

5Monitor etcd size

Track metrics and set alerts.

Advanced notes

Consider dedicated etcd instance for events.

Frequently asked questions

What does "EVENT_LIMIT_EXCEEDED" mean?
Kubernetes events can consume most of etcd storage, causing the cluster to reach quota and rejecting new writes.
How do I fix "EVENT_LIMIT_EXCEEDED"?
etcd database full with Kubernetes events. Fix by cleaning up old events, increasing quota, or using separate etcd for events.
What causes "EVENT_LIMIT_EXCEEDED"?
Too many events generated Event cleanup/TTL not configured etcd quota too small
Is the "EVENT_LIMIT_EXCEEDED" error serious?
Yes. This is a critical-severity error that usually blocks work until it is fixed, but the steps above resolve it.

Official resources & further reading