Scaling the etcd cluster with client and peer certificate authentication enabled.

[ugur99]

we'd like to ask suggested way of scaling up etcd cluster; we've been using kubespray for a while and its default approach was to restart all etcd instances at the same time after regenerating all etcd related certs; but this approach causes some downtime for etcd cluster as expected. I am not sure if this behaviour is intended; but it is.

So we'd like to be sure: when we add a new etcd member to the cluster; can we restart old instances one by one without losing quorum?

If this would cause some problems; perhaps rotating the certs of old members first and restarting them one at a time without adding new members sounds like a safer way to do this.

Maybe I'm worrying for nothing. But I just wanted to make sure. Thanks in advance for your help.

[ugur99]

sorry for my confusion; closing this issue