From nobody@FreeBSD.org  Thu Nov  7 01:03:08 2013
Return-Path: <nobody@FreeBSD.org>
Received: from mx1.freebsd.org (mx1.freebsd.org [8.8.178.115])
	(using TLSv1 with cipher ADH-AES256-SHA (256/256 bits))
	(No client certificate requested)
	by hub.freebsd.org (Postfix) with ESMTP id 891174D6
	for <freebsd-gnats-submit@FreeBSD.org>; Thu,  7 Nov 2013 01:03:08 +0000 (UTC)
	(envelope-from nobody@FreeBSD.org)
Received: from oldred.freebsd.org (oldred.freebsd.org [8.8.178.121])
	(using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits))
	(No client certificate requested)
	by mx1.freebsd.org (Postfix) with ESMTPS id 76CA92D41
	for <freebsd-gnats-submit@FreeBSD.org>; Thu,  7 Nov 2013 01:03:08 +0000 (UTC)
Received: from oldred.freebsd.org ([127.0.1.6])
	by oldred.freebsd.org (8.14.5/8.14.7) with ESMTP id rA7138jT069097
	for <freebsd-gnats-submit@FreeBSD.org>; Thu, 7 Nov 2013 01:03:08 GMT
	(envelope-from nobody@oldred.freebsd.org)
Received: (from nobody@localhost)
	by oldred.freebsd.org (8.14.5/8.14.5/Submit) id rA713828069086;
	Thu, 7 Nov 2013 01:03:08 GMT
	(envelope-from nobody)
Message-Id: <201311070103.rA713828069086@oldred.freebsd.org>
Date: Thu, 7 Nov 2013 01:03:08 GMT
From: Andrew Kinney <andykinney@advantagecom.net>
To: freebsd-gnats-submit@FreeBSD.org
Subject: igb interface output byte counter double real output bytes on 8.4-RELEASE generic kernel
X-Send-Pr-Version: www-3.1
X-GNATS-Notify:

>Number:         183732
>Category:       kern
>Synopsis:       [igb] igb interface output byte counter double real output bytes on 8.4-RELEASE generic kernel
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-net
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Nov 07 01:10:00 UTC 2013
>Closed-Date:    Sat Nov 16 08:43:13 UTC 2013
>Last-Modified:  Mon Nov 18 23:50:00 UTC 2013
>Originator:     Andrew Kinney
>Release:        8.4-RELEASE
>Organization:
Advantagecom Networks, Inc.
>Environment:
FreeBSD redact.redact.com 8.4-RELEASE-p4 FreeBSD 8.4-RELEASE-p4 #0: Mon Sep  9 23:16:13 UTC 2013     root@amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC  amd64
>Description:
With a generic unmodified 8.4-RELEASE-p4 kernel, igb interface outbound byte counters are double real output bytes. Packet counts are as expected.

This existing problem report against 8.4-STABLE seems to identify the root cause, but it was largely ignored because it wasn't reported against 8.4-RELEASE:
http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/182828

Symptoms visible on a box running bridging and IPFW with two igb interfaces. Outbound bytes is usually about double inbound bytes on the opposite interface. Packet counts are about equal. Very visible when there is significant traffic graphed via snmpd and cacti.

Diagnosed with:
netstat -I igb0 -btW ; netstat -I igb2 -btW ; date
# wait a few seconds
netstat -I igb0 -btW ; netstat -I igb2 -btW ; date
# compare Ibytes delta on one interface to Obytes delta on opposite interface

Actual Results: Obytes delta on one interface should be roughly equal to Ibytes delta on the opposite interface.

Expected Results: Ibytes on one interface should be roughly equal to Obytes on the opposite interface
>How-To-Repeat:
Use FreeBSD-8.4-RELEASE with igb interfaces. Observe Obytes from "netstat -I igb0 -btW" with a known quantity of network traffic before and after the traffic. Obytes will be approximately double the expected value.
>Fix:
patch submitted against FreeBSD-8.4-STABLE on
http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/182828

>Release-Note:
>Audit-Trail:

From: Andrew Kinney <andykinney@advantagecom.net>
To: bug-followup@FreeBSD.org, andykinney@advantagecom.net
Cc:  
Subject: Re: kern/183732: igb interface output byte counter double real output
 bytes on 8.4-RELEASE generic kernel
Date: Wed, 06 Nov 2013 17:31:32 -0800

 "Actual Results" should instead read:
 
 Actual Results: Obytes delta on one interface is roughly double Ibytes 
 delta on the opposite interface.
Responsible-Changed-From-To: freebsd-bugs->freebsd-net 
Responsible-Changed-By: linimon 
Responsible-Changed-When: Thu Nov 7 02:50:14 UTC 2013 
Responsible-Changed-Why:  
Over to maintainer(s). 

http://www.freebsd.org/cgi/query-pr.cgi?pr=183732 
State-Changed-From-To: open->closed 
State-Changed-By: glebius 
State-Changed-When: Sat Nov 16 08:28:46 UTC 2013 
State-Changed-Why:  
Andrew, 

1) the PR is a duplicate of already existing PR and you knew that. 
2) the problem doesn't exist in 9.2-RELEASE 
3) there will be no more releases cut from the stable/8 branch, 
since bug isn't critical, and isn't security related, it doesn't 
deserve Errata Notice. 

Taking the above into account, I am closing the PR. 

Please, upgrade to 9.2-RELEASE or upcoming soon 10.0-RELEASE. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=183732 

From: Andrew Kinney <andykinney@advantagecom.net>
To: bug-followup@FreeBSD.org, andykinney@advantagecom.net
Cc:  
Subject: Re: kern/183732: [igb] igb interface output byte counter double real
 output bytes on 8.4-RELEASE generic kernel
Date: Mon, 18 Nov 2013 15:21:36 -0800

 The other bug was against 8.4-stable, which receive no action and no 
 work because it isn't RELEASE. This PR is against 8.4-RELEASE.
 
 It may seem like I'm splitting hairs to say this isn't duplicate, but 
 read the follow-ups on the other bug. Nobody cares because that one is 
 not against a released version. That's why I submitted the PR against a 
 released version that someone somewhere is supposed to care about.
 
 If feel you have to close a bug, I think I've made a clear argument for 
 you to re-open this PR and close the other one.
 
 8.4-RELEASE is a supported release. You stating I should jump on 9.2 
 (and a completely new set of unknown problems for a production machine) 
 completely neglects the fact that 8.4-RELEASE is supported and this 
 should get attention.
 
 This bug needs to get fixed. While it doesn't cause crashes for prevent 
 booting, it does cause problems for billing our clients. We use the byte 
 counters for billing and costing. If it were always exactly double the 
 real bytes, we could deal with it, but it's not. Sometimes the delta in 
 bytes drops to about 1/10th the real traffic for periods of 15 to 30 
 minutes at a time. So, it's anyone's guess what the real byte count 
 should have been. Good luck ever convincing a customer with a large 
 overage that they should trust your numbers after they see this.
 
 If public policy for the FreeBSD project has changed and you only want 
 to support one release, then make that policy change public. Don't dance 
 around excusing bugs for supposedly supported releases because I'm wrong 
 for having chosen something other than the latest version.
 
 Sincerely,
 Andrew Kinney
 President / CTO
 Advantagecom Networks, Inc.
>Unformatted:
