From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rick Jones Subject: Re: bug in arp handling 2.6.20 Date: Thu, 06 Sep 2007 18:16:51 -0700 Message-ID: <46E0A683.4030902@hp.com> References: <50155211-787D-48AE-B9D8-7FD18406C8E3@ontoserve.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: Joseph Southwell Return-path: Received: from palrel11.hp.com ([156.153.255.246]:48648 "EHLO palrel11.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751654AbXIGBRX (ORCPT ); Thu, 6 Sep 2007 21:17:23 -0400 In-Reply-To: <50155211-787D-48AE-B9D8-7FD18406C8E3@ontoserve.com> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Joseph Southwell wrote: > scenario > Machine A > eth0 is plugged into network. > 192.168.1.201 > eth0:1 > 192.168.1.2 > > > Machine B > eth1 is plugged into network. > 192.168.1.101 > eth0 is not plugged into network. so I can test locally before > switching the wire. > 192.168.1.201 > eth0:1 > 192.168.1.2 > > arp request for 192.168.1.2 is responded to by eth1 on machine B. That > seems buggy to me. Does it still do so if you set either "arp_ignore" or "arp_filter" (I can never keep those two straight)? Could be the "weak-end-system" design of the stack at work here. That allows any interface in the system to respond to an ARP request for any of the IP's assigned to any of the interfaces. > worse, it continues to do it after unconfiguring > eth0:1 and eth0. As in simply marking it "down?" rick jones