alyssawilk

/wait on CI

Those checks were added because the auto_config is flat out not expected to work correctly with plaintext sockets. You can't ALPN without alpn support. I think it's largely luck it...

cc @snowp @mattklein123 @RyanTheOptimist thoughts? summary: The mixed connection pool is designed for TLS ALPN. It does "fails open" so does HTTP/1.1 if ALPN is not configured https://github.com/envoyproxy/envoy/blob/main/source/common/http/mixed_conn_pool.cc#L45 The only...

Ok I don't think there's consensus here so I'm going to lean harder on (2) and say folks can add an option for do_not_enforce_alpn which will fail open to HTTP/1.1...

I don't recall if we had an open issue (@DavidSchinazi?) and I think it's pretty far back on our queue but if this is for debugging event flow it's worth...

-> Ryan for C++ as I'm largely out this week

let's do it! It AFIK shouldn't reduce performance at all, as ALPN doesn't add lantecy. When we do this can we get rid of the base_http2 cluster entirely?

oh, I "fixed" that a while back https://www.envoyproxy.io/docs/envoy/latest/version_history/v1.23/v1.23.0#minor-behavior-changes changed HTTP/2 connection pooling and the ALPN pool to remember the number of streams allowed by the endpoint and cap multiplexed streams...

Yeah, I agree this is reasonable, but I've got a few other things in my queue so I'll leave it as help wanted for now.

no updates from me as I'm not working on it. I believe no one has picked it up still.