Hi gRPC team,

I opened a GitHub issue here: https://github.com/grpc/grpc/issues/41492

*Summary:* In a C++ server-streaming WriteReactor, if the client aborts
(grpc-web via Envoy) while a write is in-flight, OnWriteDone(false) is not
invoked until ~300s later. During this time, the stream blocks and can
stall subsequent streams on the same HTTP/2 connection.

This behavior is currently blocking multiple time-critical customer
workflows for us, so we’re trying to understand whether this is expected or
if there’s a better-supported pattern.

*We verified:*

   - StartWrite() was called
   - Client abort occurs
   - OnWriteDone(false) fires only after Envoy’s ~300s idle timeout

*Question:* Is this delayed OnWriteDone behavior expected per gRPC C++
reactor semantics, or should client cancellation trigger immediate
completion of in-flight writes? Is there another mechanism I can use to
enable OnWriteDone failure sooner?

Thanks — happy to provide logs or a minimal repro if useful.

*Ajit Singh*

Software Engineer

[email protected] <[email protected]>

(856) 533-0616

Applied Intuition, Inc. <https://applied.co>

-- 
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 [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/grpc-io/CADfWbLh7XP5HXXJHkTF3v5Ucuh6kZGXb4%2Buk%2B1rPYt_bH4xscQ%40mail.gmail.com.

Reply via email to