According to this blog post, Resin reuses VPN tunnels when forwarding web requests to devices.
Can you share what the timeout is (if any) for these persisted tunnels? And are there any plans to make this configurable?
Hey, as I understand it the tunnels persist for 65s of inactivity, so will be reused as long as the next request comes in shortly after the first one terminates. I remember when we tested it there were huge benefits for http traffic which regularly establishes/reestablishes connections, however for long running connections it makes much less difference as they spend a lot less time on establishing connections.
For customising this timeout, there’s no plan for it that I’m aware of as the majority of the benefits come from patterns that establish/reestablish a bunch of connections in a short period, and I believe that is covered by the current timeout - if you have a use case that it would make sense for though then I’d love to hear as I know our vpn performance is something @wrboyce has put quite a lot of work into