HA configuration is not HA

From: http://docs.rancher.com/rancher/installing-rancher/installing-server/multi-nodes/
“The IP that you configure for the websocket proxy needs be accessible from your browser and from the hosts.”

And websocket proxy is JUST ONE instance and it is Single point of failure! Or am I wrong? What will happen to hosts who fail to reach websocket-proxy?

Edit: Sorry, I have read it whole now. There is paragraph about that websocket-proxy is not HA. Is there some ETA to solve this? Question about hosts who cannot reach websocket-proxy is essential.

We are close to releasing a more complete HA solution. There is always a danger to giving out exact dates on a public forum, so I’ll say you can expect something inside of two weeks. But, things can change :slight_smile:

I believe websocket-proxy will be managed in a active/passive setup. So if the active one goes down, the passive one will take over. Hosts that cannot reach websocket-proxy will go into a “Reconnecting” state. They will continually try to reconnect to it and once they do, they will return to a normal “Active” state. While in the reconnecting state, no new containers will be scheduled to the host but existing containers will ocntinue to run.

1 Like