My environment looks like this:
Two datacenters, each datacenter running 7 Docker hosts, datacenters are connected via a VPN tunnel, Rancher Server lives in one of the datacenters, each datacenter is setup as an “Environment”.
Everything works great when the VPN tunnel is up, however when the VPN tunnel goes down everything in the datacenter where the Rancher Server doesn’t live becomes unavailable because the load balancer, which is deployed to specific hosts, starts throwing 503 errors. Things instantly start working again when the Rancher server becomes available again.
So I’m curious, is the Rancher Server a single point of failure for using the Rancher Load Balancer service? I’m a bit surprised that losing the rancher server instantly makes the Load Balancer service forget about all of it’s previously available containers and instead has it start tossing 503’s.