From bruce@insane.noc.clara.net  Fri Aug 30 08:13:51 2002
Return-Path: <bruce@insane.noc.clara.net>
Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 7DD2437B400
	for <FreeBSD-gnats-submit@freebsd.org>; Fri, 30 Aug 2002 08:13:51 -0700 (PDT)
Received: from insane.noc.clara.net (insane.noc.clara.net [195.8.70.97])
	by mx1.FreeBSD.org (Postfix) with ESMTP id F35D343E3B
	for <FreeBSD-gnats-submit@freebsd.org>; Fri, 30 Aug 2002 08:13:50 -0700 (PDT)
	(envelope-from bruce@insane.noc.clara.net)
Received: from bruce by insane.noc.clara.net with local (Exim 3.34 #1)
	id 17knT3-000NP7-00
	for FreeBSD-gnats-submit@freebsd.org; Fri, 30 Aug 2002 16:13:49 +0100
Message-Id: <E17knT3-000NP7-00@insane.noc.clara.net>
Date: Fri, 30 Aug 2002 16:13:49 +0100
From: Bruce Dixie <bruce@uk.clara.net>
Sender: Bruce Dixie <bruce@insane.noc.clara.net>
Reply-To: Bruce Dixie <bruce@uk.clara.net>
To: FreeBSD-gnats-submit@freebsd.org
Cc:
Subject: simultaneous multiple server network failure
X-Send-Pr-Version: 3.113
X-GNATS-Notify:

>Number:         42216
>Category:       kern
>Synopsis:       [fxp] simultaneous multiple server network failure
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    rwatson
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Fri Aug 30 08:20:01 PDT 2002
>Closed-Date:    Sun Jan 13 15:19:41 UTC 2008
>Last-Modified:  Sun Jan 13 15:19:41 UTC 2008
>Originator:     Bruce Dixie
>Release:        FreeBSD 4.6-STABLE i386
>Organization:
>Environment:
System: Multiple systems ranging from 4.5R to 4.6Stable
>Description:
Within 10 seconds of each other, 8 FreeBSD servers lost connectivity. Linux and Solaris servers on the same network and performing the same tasks were unaffected. The only log in all the servers /var/log/messages is 'Aug 29 01:28:52 eos /kernel: fxp0: device timeout' - this appears in all the servers logs. The time varies for each server by no more than 10secs. As soon as we disconnected the server we suspected of causing the problem connectivity returned. There's nothing in the logs of the suspect server , but it was unresponsive to the keyboard. As soon as the server was disconnected all the other servers started responding - no reboot needed. I'm submitting this as suggested by Mark Murrey.
>How-To-Repeat:
	
>Fix:

	


>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->patched 
State-Changed-By: kris 
State-Changed-When: Mon Jul 14 03:34:21 PDT 2003 
State-Changed-Why:  
This is likely to be caused by an fxp card's packet buffer 
filling while the system is unable to process them, causing 
the card to go into a loop transmitting ethernet quench 
packets and flooding the ethernet segment.  See 

http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/dev/fxp/if_fxp.c?rev=1.179&content-type=text/x-cvsweb-markup 

for a description of the problem and the code committed to 5.1. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=42216 
Responsible-Changed-From-To: freebsd-bugs->rwatson 
Responsible-Changed-By: kris 
Responsible-Changed-When: Mon Jul 14 03:37:41 PDT 2003 
Responsible-Changed-Why:  
Assign to rwatson to consider MFCing the workaround 

http://www.freebsd.org/cgi/query-pr.cgi?pr=42216 
State-Changed-From-To: patched->feedback 
State-Changed-By: rwatson 
State-Changed-When: Sat Jan 17 15:04:49 PST 2004 
State-Changed-Why:  
I've merged the tunable as if_fxp.c:1.110.2.32; please let me know 
if it doesn't work as expected, or if there are any problems. 


http://www.freebsd.org/cgi/query-pr.cgi?pr=42216 
State-Changed-From-To: feedback->suspended 
State-Changed-By: rwatson 
State-Changed-When: Sun May 2 15:28:50 PDT 2004 
State-Changed-Why:  
Suspend PR rather than close it as I think we should probably change 
the default to be "flow control disabled" rather than "flow control 
enabled" due to the potential seriousness of issues resulting from 
this bug being triggered. 


http://www.freebsd.org/cgi/query-pr.cgi?pr=42216 
State-Changed-From-To: suspended->closed 
State-Changed-By: rwatson 
State-Changed-When: Sun Jan 13 15:17:57 UTC 2008 
State-Changed-Why:  
Close PR: I disabled flow control in fxp by default in 2005, so the poor 
default issue is now corrected. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=42216 
>Unformatted:
