From owner-freebsd-net@FreeBSD.ORG Fri Oct 29 17:04:38 2010 Return-Path: Delivered-To: freebsd-net@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 8660F1065672 for ; Fri, 29 Oct 2010 17:04:38 +0000 (UTC) (envelope-from tom@tomjudge.com) Received: from eu1sys200aog102.obsmtp.com (eu1sys200aog102.obsmtp.com [207.126.144.113]) by mx1.freebsd.org (Postfix) with SMTP id 8DE5C8FC0A for ; Fri, 29 Oct 2010 17:04:36 +0000 (UTC) Received: from source ([63.174.175.251]) by eu1sys200aob102.postini.com ([207.126.147.11]) with SMTP ID DSNKTMr+o7tw8Nplz0jCTDNXJy7Er68wp4es@postini.com; Fri, 29 Oct 2010 17:04:37 UTC Received: from [172.17.10.53] (unknown [172.17.10.53]) by bbbx3.usdmm.com (Postfix) with ESMTP id BDE3BFD063; Fri, 29 Oct 2010 17:04:34 +0000 (UTC) Message-ID: <4CCAFE85.10302@tomjudge.com> Date: Fri, 29 Oct 2010 12:04:05 -0500 From: Tom Judge User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.11) Gecko/20101006 Lightning/1.0b2 Thunderbird/3.1.5 MIME-Version: 1.0 To: sbruno@freebsd.org References: <1288204022.3808.191.camel@home-yahoo> In-Reply-To: <1288204022.3808.191.camel@home-yahoo> X-Enigmail-Version: 1.1.1 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: "freebsd-net@freebsd.org" , Sean Bruno Subject: Re: Excessive Duplicate ACKs X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 29 Oct 2010 17:04:38 -0000 On 10/27/2010 01:27 PM, Sean Bruno wrote: > We moved an application stack from BSD4(BOO!) to BSD7(YAY!) recently and > got a great performance increase, so first: GOOD JOB. > > Periodically, we are seeing strings of duplicate ACK being sent in > <100uSec deltas. I can't imagine that this should be happening, but > there it is. I've sanitized an example trace of a transaction, > demonstrating an average case that had 8 dup ACKs, some less than 3 > microsends in delta! sysctl -a also attached (and sanitized). > I saw something like this on FreeBSD 7.1 with an oldish slapd where the daemon was sending very small frames very quickly and the receiver could not keep up and was dropping the frames in the NIC firmware has the host was not draining the receive buffer fast enough. We used wireshark to analyze the flows and it was reporting lots of duplicate ACKs in the capture. Looking more closely at the ACK we saw that all of the reported duplicates where actually SACK frames and wireshark was not smart enough to look into the SACK option and was only processing the ACK sequence number. Later in the flow (microseconds also) we where seeing a fast retransmission of the data between the current ACK sequence number and the SACK start sequence. Maybe you are seeing something similar? Tom > http://people.freebsd.org/~sbruno/dup_ack_collapsed.txt > > http://people.freebsd.org/~sbruno/sysctl_dup_acks.txt > > > Sean > > _______________________________________________ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org" -- TJU13-ARIN