I have not done a ton of failover stuff so I am trying to verify if you can failover to a full-blown gateway.
If I have Gateway A and Gateway B and B goes down, can I failover my clients to the A side if there is a project of the same name even though it is not redundant. Instead, I have it setup as Independent.
Local client fallback is essentially a hard cutover to whatever the local gateway designates as the fallback project:
There's no attempt to transfer client state, and no need for the project names or anything to line up. It's just a more-or-less "seamless" [1] switch from running one project from one gateway to running a different project from the local gateway (akin to a system.util.retarget operation).
how seamless it will be depends on how closely aligned the two versions are; switching versions may require a JVM restart which adds time, but it will still be automatic ↩︎
Is there a way to achieve this for Perspective as well?
In workstation, yes:
Not for browser launched sessions. I'm not sure about the mobile app...
For a web browser I theory you could detect that the gateway is unreachable and then redirect to a new URL of your "failover". It would just not be quite seamless.
You could, using some kind of loadbalancing strategy or some other service "in front of" Ignition. We cannot, because if you point at the gateway's IP (while it's down), your web browser is just not going to find anything. Once a session is launched, we do cache some info and attempt to seamlessly switch over to the backup node, for redundancy failover, but we don't have any analogue for falling back to arbitrary different gateways. And we have no interest in maintaining/providing a load balancer or reverse proxy first party, because that's well tread ground that you can easily find free solutions for.