Re: HTTP/2 response completed before its request

• From: Martin Thomson <martin.thomson@gmail.com>
• Date: Tue, 1 Jul 2014 11:54:08 -0700
• To: William Chan (陈智昌) <willchan@chromium.org>
• Cc: Jesse Wilson <jesse@swank.ca>, HTTP Working Group <ietf-http-wg@w3.org>
• Message-ID: <CABkgnnVku6o+4k5wiJ3DZ6=4mMGf=yN5K2bX1voD0uuugJw66Q@mail.gmail.com>

On 1 July 2014 11:48, William Chan (陈智昌) <willchan@chromium.org> wrote:
> I think Jesse didn't explicitly say that the response didn't complete. The
> application code at the client that issued the request hasn't started
> reading the response. It's not pumping both the write and reads, just the
> writes. Therefore, its receive window can shrink to 0 and stall the server's
> sending of the response. Both the request and responses are stalled on flow
> control.


Yes, but the fix is to pump reads too.  And if the server finishes the
response (i.e., we have a transition to half closed (remote) at the
client), then the client has no reason to continue to waste uplink
bandwidth.  So RST_STREAM NO_ERROR seems about right.

Received on Tuesday, 1 July 2014 18:54:35 UTC