From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ie0-f174.google.com (mail-ie0-f174.google.com [209.85.223.174]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id 545BE208AAD for ; Sun, 6 Jan 2013 16:22:26 -0800 (PST) Received: by mail-ie0-f174.google.com with SMTP id c11so22543619ieb.33 for ; Sun, 06 Jan 2013 16:22:25 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=/G/GW4RM0Clhnx5VX5f/xnEwdVwjMYRiYLzjpeN9d3M=; b=TxexvAKTxD8Gi1d5/ZJvSp83Z62WfP7eBeMRTeocbvhznydj5nWvXWrQeEPSo5RVI2 1O3UOk9Pt1gcknm7egnCYrHpeb04ipI4OzAjObIx5Lzm6s75asz9mAxMkZCGemTF6mKD 3+7LJ0d7sOXEBaOQGLBRRCPxDU7VM/r9N8G3L2FFRkcapvkF5SfYy1IAGDFLEhbAUTrB 4g/vVmdzup+O5+iCbO6mg+tzJEia+4jRsiLVgxXWxqaG97BATHKTK+JnUDEoTCyizFR/ O7Xchmy/FA0+YtkOIYYtZVV9s7Bhg6SWbynd62K0Xeai5rrLE+dtubeFTLwIeASvheDs ffCg== MIME-Version: 1.0 Received: by 10.50.88.136 with SMTP id bg8mr4350005igb.96.1357518145461; Sun, 06 Jan 2013 16:22:25 -0800 (PST) Received: by 10.64.135.39 with HTTP; Sun, 6 Jan 2013 16:22:25 -0800 (PST) Date: Mon, 7 Jan 2013 00:22:25 +0000 Message-ID: From: Dave Taht To: cerowrt-devel@lists.bufferbloat.net Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Subject: [Cerowrt-devel] bug in simple_qos.sh on ingress that has been there for a long time... X-BeenThere: cerowrt-devel@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: Development issues regarding the cerowrt test router project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 07 Jan 2013 00:22:26 -0000 I have been puzzled over some classification related behaviors for quite some time. I just finally figured out that the simple_qos script is buggy on ingress, in that it dumps most of the traffic into the "background" bin. Things like netperf -Y CS1,CS1 -H snapon.lab.bufferbloat.net # ends up in background bin (1:13) netperf -Y CS0,CS0 -H snapon.lab.bufferbloat.net # ALSO ends up in background bin (1:13) tc -s qdisc show dev ifb0 So there is a bug in the "diffserv" component of that script in that this match matches almost everything for some reason. The offending line is: tc filter add dev $DEV protocol ip parent 1:0 prio 6 u32 match ip tos 0x20 0xfc classid 1:13 # CS1 Bulk On egress it uses the more "sane to express" iptables version (which won't work on ingress) to do similar classification. The reason for the 0xfc was to filter out ECN bits. Maybe I've got that backwards? or these filters need to be differently masked? or? /me has brain fail to do an and Anyway, for those collecting data on up/down bandwidths and playing with classification, this bug is going to mess up your world. It just did mine, anyway. --=20 Dave T=E4ht Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.= html