Allow configuring max_local_error_reset_streams#2421
Allow configuring max_local_error_reset_streams#2421tottoto merged 4 commits intohyperium:v0.14.xfrom
Conversation
|
Given there's now also references from h2, would it be possible to get this merged? (cc @tottoto perhaps? thanks) |
|
I'm not sure why the "CI / Interop Tests (macOS-latest)" test failed, but it looks unrelated? |
tottoto
left a comment
tottoto
left a comment
There was a problem hiding this comment.
@alexanderrilee Thanks! Could you squash the commits and upstream it to the default branch? I think I'd like to merge this pull request as a backport of it.
I'm not sure why the "CI / Interop Tests (macOS-latest)" test failed, but it looks unrelated?
Yes. It isn't related to this change.
alexanderrilee
force-pushed
the
configure-max-stream-resets
branch
from
7c3f006 to
3aba7c5
Compare
…2437) Squashed version of #2421 applied to `master`. Co-authored-by: alexanderrilee <[email protected]>
|
@tottoto, do you know if we can bump the minor version to capture this change? |
|
@LucioFranco, maybe you could help with the minor version bump? |
|
@alexanderrilee yeah, I will work on a patch release 👍 |
|
Hey, @LucioFranco, just wanted to check on the minor version release 🙏 |
4 participants
Motivation
This makes max_local_error_reset_streams configurable. We have an isolated environment where we create and close many (valid) short-lived gRPC streams. We very quickly hit the 1024 limit on stream resets, which initiates a GoAway. During connection recreation, the Python gRPC client misbehaves. We would like the option to tune this limit, since this is a safe traffic pattern for us.
Solution
I've added
http2_max_local_error_reset_streamsto theServerAPI.