From owner-freebsd-stable Thu Oct 3 5:48:38 2002 Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5DCE337B401 for ; Thu, 3 Oct 2002 05:48:36 -0700 (PDT) Received: from anuket.mj.niksun.com (gateway.niksun.com [65.115.46.162]) by mx1.FreeBSD.org (Postfix) with ESMTP id 6FE0043E3B for ; Thu, 3 Oct 2002 05:48:35 -0700 (PDT) (envelope-from ath@niksun.com) Received: from stiegl.mj.niksun.com (stiegl.mj.niksun.com [10.70.0.231]) by anuket.mj.niksun.com (8.12.3/8.12.3) with ESMTP id g93ClavD062467; Thu, 3 Oct 2002 08:47:36 -0400 (EDT) (envelope-from ath@stiegl.mj.niksun.com) X-RAV-AntiVirus: This e-mail has been scanned for viruses. Received: (from ath@localhost) by stiegl.mj.niksun.com (8.11.6/8.11.6) id g93ClbL24962; Thu, 3 Oct 2002 08:47:37 -0400 (EDT) (envelope-from ath@stiegl.mj.niksun.com) To: Tony Maher Cc: stable@FreeBSD.ORG Subject: Re: 'losing' every second packet References: <200210022355.g92NtU1o029952@dt.home> From: Andrew Heybey Date: 03 Oct 2002 08:47:37 -0400 In-Reply-To: <200210022355.g92NtU1o029952@dt.home> Message-ID: <853crnvgd2.fsf@stiegl.mj.niksun.com> Lines: 73 User-Agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.1 (Cuyahoga Valley) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG > about 3-4 weeks ago my home firewall started to exhibit a strange > behaviour of losing what appears to be every second packet but _only_ > after it has been running for more than 12 hrs (typically around the > 24hr mark). The only way to 'fix' it is to reboot. Restarting firewall > rules does not help. > It was also about this time I modified my firewall to use squid and > started to run darkstat on both internal and external card. So I > initially assumed it was something to do with that. However I have > since turned off darkstat but the behaviour still remains. I haven't > tried turning off squid yet but I doubt that it is the problem. > I also upgrade to 4-stable a few days ago (and started using ipfw2) > but the behaviour was similar. Had to reboot after about 24 hrs for the > first couple of days but yesterday had been running for more than 72 hrs > without problems but this morning (now that I am at work) it has > reappeared. Cannot ssh to the firewall and ping shows 50% packet loss. > (FreeBSD gw 4.7-RC FreeBSD 4.7-RC #0: Wed Sep 25 06:10:10 EST 2002) > 409 packets transmitted, 207 packets received, 49% packet loss > round-trip min/avg/max/stddev = 30.449/35.600/75.392/6.487 ms > (was also trying to ssh at the time so not 50% exactly) > Yesterday I also got confirmation from a colleague who runs a freebsd > firewall as well that he has similar problems (large packet loss > around 50%) in the same time-scale. He is unlikely to have modified > anything since we set it up. (He is away at present so I cannot get any > definitive answers about his setup. I do not think it has been upgraded > for several months, so it is probably release version 4.4). > We also run a freebsd firewall at work 4.4-STABLE that works perfectly > (uptime 220+ days). > What is in common is that our home firewalls are conected via cable > modem to optusnet.com.au (and the work firewall is on entirely > different network). Also about 4 weeks ago optus changed their > charging policy to a 3Gb limit and maybe other changes to the > network?! Probes?! > Seems to be more than a coincidence. > Anybody else experiencing these problems? I have seen almost exactly this behavior on my cable modem in the US. "arp -ad" fixes the problem. I have the following /usr/local/etc/rc.d/arp_delete.sh: #!/bin/sh (while /usr/bin/true; do sleep 3600 /usr/sbin/arp -ad done) & Kind of tasteless, but it does the job. Since I figured out the above work-around, I have noticed log messages like the following: Sep 17 13:40:17 celis /kernel: arplookup 10.150.16.1 failed: host is not on local network Combined with the following (from /var/db/dhclient.leases): option dhcp-server-identifier 10.150.1.2 and since I do NAT with my internal network on net 10, I suspect that there is some problem with the dhcp server or router not hearing from me often enough and somehow forgetting about me. Regardless, deleting the arp cache periodcially fixes the problem. andrew To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message