- From: Martin Thomson <martin.thomson@gmail.com>
- Date: 2014年11月24日 21:09:21 +0000
- To: Michael Sweet <msweet@apple.com>, "brad@danga.com" <brad@danga.com>
- Cc: Willy Tarreau <w@1wt.eu>, Roberto Peon <grmocg@gmail.com>, Matthew Kerwin <matthew@kerwin.net.au>, HTTP Working Group <ietf-http-wg@w3.org>
- Message-ID: <CABkgnnUt0bSZ9Yz5Gf5WF=M8wxjniLH8uD2yLLJtYYp++8w0xA@mail.gmail.com>
That is a separate problem. We are only talking about the violation of settings here. You are right that this is a general problem, but not a new one. This is partly why we encourage large concurrency limits, and only use SHOULD when talking about the single connection. On Mon, Nov 24, 2014, 01:46 Michael Sweet <msweet@apple.com> wrote: What about proxies? They can advertise a limit but the upstream server may have different limits. Short of opening extra connections to support the required amount of concurrency, seems like it should be able to tell the client it has reached a limit and to try again later...
Received on Monday, 24 November 2014 21:09:48 UTC