Hey! I did, also I went through the logs from the secure_endpoint component.
I've added a counter to each message buffer, so I could follow each message in the trace logs. It seems just before the client stops, it sends a (first part of) request to the server side (which is not in the server side logs) and receives a couple responses from previous requests. Although the last message never reached the server. Logs are here: https://mega.nz/file/FxJExLiZ#YlTwpIx_FadyRtcT5DDmz4pCefDxeBsixkADGcUJJpo Thanks, Peter Yuri Golobokov <golobo...@google.com> ezt írta (időpont: 2024. jún. 10., H, 18:50): > Did you try just the http, i.e., `GRPC_TRACE=http` ? > > On Mon, Jun 10, 2024 at 3:43 AM Peter Olah <p.nothin...@gmail.com> wrote: > >> Hey! >> >> Thanks for the suggestion. I tried to compare those logs, but I'm not >> sure what to look for. The trace log contains an overwhelming amount of log >> lines, I can hardly mark the boundaries between each message. Do you have >> any suggestion what component could be excluded from the trace? Are there >> any troubleshooting guides? >> >> Thanks, >> Peter >> >> Yuri Golobokov <golobo...@google.com> ezt írta (időpont: 2024. jún. 7., >> P, 20:56): >> >>> Have you tried capturing debug logs on the client and the server? I >>> think comparing the logs of CF path vs the logs of direct connection might >>> help. >>> >>> On Fri, Jun 7, 2024 at 11:28 AM Peter Olah <p.nothin...@gmail.com> >>> wrote: >>> >>>> Hey, >>>> >>>> We're facing issues when we're trying to communicate via a gRPC >>>> endpoint where both the input and return values are streams (bidirectional >>>> streaming). In the attached minimal example, we're having a service which >>>> hashes the received data upon requests and returns with calculated hash. By >>>> increasing the data size in the client the communication becomes more >>>> unreliable. The sweet spot seems to be somewhere between 32k and 64k. >>>> >>>> However we don't understand the reason for this behaviour. We didn't >>>> get any real error, the communication just 'stops'. Eventually it times out >>>> and the channel gets torn down. We're experiencing this behaviour only when >>>> we're communicating through CF: in case of a direct connection to one of >>>> our IP it doesn't happen. >>>> >>>> Please help us understand what's happening here. >>>> >>>> The minimal example can be found here: >>>> https://github.com/nothingam/grpc-hash-service >>>> >>>> Thanks, >>>> Peter >>>> >>>> -- >>>> You received this message because you are subscribed to the Google >>>> Groups "grpc.io" group. >>>> To unsubscribe from this group and stop receiving emails from it, send >>>> an email to grpc-io+unsubscr...@googlegroups.com. >>>> To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/grpc-io/CAJDqnpLWr%3D1MLQvjiJcG85NkhV%2B82rpRSof7wZqYU5jnoCxjig%40mail.gmail.com >>>> <https://groups.google.com/d/msgid/grpc-io/CAJDqnpLWr%3D1MLQvjiJcG85NkhV%2B82rpRSof7wZqYU5jnoCxjig%40mail.gmail.com?utm_medium=email&utm_source=footer> >>>> . >>>> >>> -- You received this message because you are subscribed to the Google Groups "grpc.io" group. To unsubscribe from this group and stop receiving emails from it, send an email to grpc-io+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/grpc-io/CAJDqnp%2BfUpAh%3Dkhevxz6rpnV8zrFRQKSpn%3Dp-T0d%2BjZ4tgX4FA%40mail.gmail.com.