Kubernetes etcd Issues Recipe

1. Search etcd logs (/var/logs/pods/*etcd*/etcd/*):
   1. leader failed to send out heartbeat on time; took too long, leader is overloaded likely from slow disk
   2. Compactions greater than 100ms

      grep -r "finished scheduled compaction" /var/log/pods/*etcd*/etcd/* | awk -F\" '$(NF-1) ~ /[0-9]s/ || $(NF-1) ~ /[1-9][0-9][0-9]/ {print $(NF-1);}' | sort -nr | head

2. Consider if defragmentation is necessary which may reduce memory usage of kube-apiserver and etcd:
   1. Kubernetes defragmentation documentation
   2. OpenShift defragmentation documentation

Previous Section (Kubernetes Basics Recipe) | Next Section (Kubernetes Modify Container Command) | Back to Table of Contents