What are you doing to redirect perspective browser clients to the backup gateway when the primary fails?
Are you asking how Perspective itself handles the failover?
Basically - there’s only one “real” page fetch when you launch a Perspective session - when you initially connect and we fetch all the necessary resources. During that process (and also later, while the connection to the master is good) the live frontend session automatically retrieves a list of possible addresses to try to reach the backup gateway on.
If the frontend session loses its connection to the master, it “knows” to start attempting to reach the backup gateway at any of those addresses.
What happens if the gateway is down and then the user tries to connect? They just need to know to connect to the failover gateway directly? (I have always assumed this was the answer, but honestly never tested it)
I know that a load balancer would resolve this, but in the case that isn’t present and its just a redundant set of gateways.
This is what I’m wondering. I see the diagrams with the load balancers and wonder if it is required or if Perspective will use the backup gateway automatically.
If the client uses Perspective Workstation, it will automatically try the backup as the launcher knows the backup configuration.
If a session has never connected, then connections from a browser will always fail - if you try to reach gateway A, and it’s not there, there’s nothing for Ignition to do - you’ve specified an invalid address. As @deon.korb points out, our launcher applications can do something, by pre-caching the backup address to try as a fallback.
A load balancer could be required if you expect fresh connections to be made from web browsers during times where the master is not available.