Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

x/net/http2: access to maxConcurrentStreams #17265

nathany opened this issue Sep 28, 2016 · 3 comments

x/net/http2: access to maxConcurrentStreams #17265

nathany opened this issue Sep 28, 2016 · 3 comments
help wanted Suggested Issues that may be good for new contributors looking for work to do.


Copy link

nathany commented Sep 28, 2016

What version of Go are you using (go version)?

go version go1.7.1 darwin/amd64

What operating system and processor architecture are you using (go env)?


What did you do?

I am working on an HTTP/2 client for Apple's push notification service. When using JWT tokens to authenticate, Apple initially sends maxConcurrentStreams = 1. Once authentication is complete they send maxConcurrentStreams = 500 (though this number may vary).

Currently I have a worker pool to send multiple requests concurrently. If I could monitor the maxConcurrentStreams, I would know how big the worker pool should be -- or how many workers should be active.

What did you expect to see?

Access to maxConcurrentStreams via httptrace or some other x/net/http2 API.

Possibly related: #13774

What did you see instead?

If I send more requests than maxConcurrentStreams dictates

With Go 1.7.1 http.DefaultClient I am getting the following error.

http2: server sent GOAWAY and closed the connection; LastStreamID=1, ErrCode=REFUSED_STREAM, 
debug="Maximum active streams violated for this endpoint."

With the latest x/net/http2 I'm getting slightly different errors (still looking into these).

2016/09/28 10:14:43 (9) device: [my-token], error: Post[my-token]: 
http2: no cached connection was available


2016/09/28 10:14:43 (11) device: [my-token], 
error: invalid character 'M' looking for beginning of value

In any case, the best I can do is prime the connection with a single request, and then send some number of concurrent requests, but not really knowing how many I can send.

Copy link

Doesn't really fit with httptrace as it's a connection-level setting and not a per-Request thing.

But I suppose we could add something to http2.ClientConn.

I'd entertain a CL if somebody has an idea and time.

@bradfitz bradfitz added this to the Unreleased milestone Sep 29, 2016
@bradfitz bradfitz added Suggested Issues that may be good for new contributors looking for work to do. help wanted labels Sep 29, 2016
Copy link
Contributor Author

nathany commented Sep 29, 2016

Would it be possible to notify the client (caller) when maxConcurrentStreams changes? (callback function, event channel?)

Or should this just expose the value of maxConcurrentStreams through http2.ClientConn -- with the client polling the value?

I'm also curious what the direction is for #13774 (blocking the transport)? By itself, that may prevent the REFUSED_STREAM errors I'm seeing. Though my code would be left guessing at the optimal number of concurrent requests without visibility into maxConcurrentStreams.

Copy link

Change mentions this issue: http2: use callback to expose SETTINGS updates from the server

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
help wanted Suggested Issues that may be good for new contributors looking for work to do.
None yet

No branches or pull requests

3 participants