Our etcd is taking 10-11 G memory. This is common across 4 clusters created by rancher 2.2.1. We have to reboot all servers daily for them to be responsive. Is this a known bug?