Ryan Hamilton

> > What if we only used the new 22.04 image for gcc CI, but use the old imagine for everything else? Would that help? > > with current setup...

"wasm_fail_stream" seems to be called when the wasm VM fails, fwiw. CC: @mpwarres @lizan

/assign @KBaichoo

You could certainly implement a filter to do this. If you'd like to contribute it to Envoy you can find more information about how to contribute here: https://github.com/envoyproxy/envoy/blob/main/CONTRIBUTING.md

cc: @ravenblackx @mattklein123

CC: @wbpcode @jmarantz

CC: @lizan Are you hitting this issue with HTTP/1 to Envoy? If so it sounds like you might be hitting the browser's socket limit. https://docs.diffusiondata.com/cloud/latest/manual/html/designguide/solution/support/connection_limitations.html

Ah! Ok, if you're using HTTP/1 from Chrome to Envoy, then that explains it. Chrome limits the number of connections to a given server to 6. Can you use HTTP/2...

cc: @htuch @mattklein123 It looks like the eds cluster is currently unowned according to CODEOWNERS.md: `/*/extensions/clusters/eds/ @UNOWNED @UNOWNED` But perhaps that just needs to be updated?

cc: @Amila-Rukshan @soulxu