W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2014

Re: GOAWAY and proxies

From: Roberto Peon <grmocg@gmail.com>
Date: Thu, 10 Apr 2014 16:50:50 -0700
Message-ID: <CAP+FsNcSvw-y_aqU3N5=kqvV_gQTmEqkXUL0Tg__oGvRLmLu+A@mail.gmail.com>
To: Daniel Sommermann <dcsommer@fb.com>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
I'll note that REFUSED_STREAM doesn't actually solve the problem entirely,
since it does require the user (not just the proxy) to retry.

It is hoped that alt-svc (potentially to the same name) will give us the
ability to have the client (proxy or not) reconnect to a new server (and
thus there should be no need for a LAME_DUCK given that mechanism).

On Thu, Apr 10, 2014 at 4:30 PM, Daniel Sommermann <dcsommer@fb.com> wrote:

>  Can't the proxy just issue RST_STREAM toward its clients for all the
>> affected requests if it can't tolerate the memory hit?  REFUSED_STREAM
>> seems like the right code there.
> It just occurred to me that this solution will only work if the client is
> speaking HTTP/2 to the proxy. This solution does not work for HTTP/1.1
> clients connected to a proxy which speaks HTTP/2 to the server.
> Seems like to fix this case you would either need something like the
> last_stream_id GOAWAY hack on the server side or buffering in the proxy.
Received on Thursday, 10 April 2014 23:51:17 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:14:29 UTC