Re: WGLC: p1 MUST NOT pipeline until connection is persistent

Alex Rousskov <> Wed, 01 May 2013 07:35 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 9C5B621F8DA6 for <>; Wed, 1 May 2013 00:35:46 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -10.569
X-Spam-Status: No, score=-10.569 tagged_above=-999 required=5 tests=[AWL=0.030, BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id liVRntKkRJ4H for <>; Wed, 1 May 2013 00:35:40 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 66E5D21F8E46 for <>; Wed, 1 May 2013 00:35:40 -0700 (PDT)
Received: from lists by with local (Exim 4.72) (envelope-from <>) id 1UXRZN-0001oM-KU for; Wed, 01 May 2013 07:35:17 +0000
Resent-Date: Wed, 01 May 2013 07:35:17 +0000
Resent-Message-Id: <>
Received: from ([]) by with esmtp (Exim 4.72) (envelope-from <>) id 1UXRZD-0001nS-VV for; Wed, 01 May 2013 07:35:08 +0000
Received: from ([]) by with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.72) (envelope-from <>) id 1UXRZC-0007V8-Sb for; Wed, 01 May 2013 07:35:07 +0000
Received: from [] (localhost []) (authenticated bits=0) by (8.14.3/8.14.3) with ESMTP id r417YKxI083726 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Wed, 1 May 2013 01:34:21 -0600 (MDT) (envelope-from
Message-ID: <>
Date: Wed, 01 May 2013 01:34:12 -0600
From: Alex Rousskov <>
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130329 Thunderbird/17.0.5
MIME-Version: 1.0
CC: Willy Tarreau <>
References: <> <> <> <> <>
In-Reply-To: <>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Received-SPF: pass client-ip=;;
X-W3C-Hub-Spam-Status: No, score=-3.5
X-W3C-Hub-Spam-Report: AWL=0.980, BAYES_00=-1.9, RP_MATCHES_RCVD=-2.57, SPF_HELO_PASS=-0.001, SPF_PASS=-0.001
X-W3C-Scan-Sig: 1UXRZC-0007V8-Sb aaf10aa671ac84a40be8f856cff31960
Subject: Re: WGLC: p1 MUST NOT pipeline until connection is persistent
Archived-At: <>
X-Mailing-List: <> archive/latest/17752
Precedence: list
List-Id: <>
List-Help: <>
List-Post: <>
List-Unsubscribe: <>

On 05/01/2013 12:52 AM, Willy Tarreau wrote:
> On Tue, Apr 30, 2013 at 05:22:39PM -0600, Alex Rousskov wrote:

>>>>> A client [...] MUST NOT pipeline on a retry connection until it
>>>>> knows the connection is persistent.

> I think that it was the reason why it's suggested somewhere that pipelining
> should only be done on connections that we know are persistent (ie after a
> first response)

That "somewhere" may be the above MUST NOT requirement we are discussing
on this thread :-).

>> Let's start with (a): To know that a connection "is persistent", the
>> client has to receive a response on that connection. This implies that
>> the client MUST NOT retry a failed pipelined request on a connection it
>> just opened.

I know what I was missing: The client may send a retry request on a new
connection. The above MUST NOT prohibits _pipelining_ on a new
connection, but does not prohibit sending a single request on that
connection. That resolves my concern about stuck clients.

I am still not sure why we are prohibiting retry pipelining on new
connections though. Why do we have to reuse an old connection if we want
to retry a failed pipeline?

Thank you,