ETCD logs say "database space exceeded"
Scenario
Anka VMs error or disappear and you see in the Controller logs E1108 23:09:28.804391 1 queue_server.go:184] Server Internal error: etcdserver: mvcc: database space exceeded.. There was plenty of available space on disk. Running etcdctl compact $(ETCDCTL_API=3 etcdctl --endpoints=:2379 endpoint status --write-out="json" | egrep -o '"revision":[0-9]*' | egrep -o '[0-9].*') (INSIDE OF THE ETCD CONTAINER), etcdctl debug --command-timeout=20s, followed by etcdctl alarm disarm fixes it, but you’d like to prevent this in the future.
Common Causes
- Lots of VMS, Nodes, and tasks
Solution
Increase the space quota from the default 2GB: https://etcd.io/docs/v3.4/op-guide/maintenance/#space-quota