From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: Possible networking regression in 3.6.0 Date: Mon, 01 Oct 2012 17:31:38 +0200 Message-ID: <1349105498.12401.706.camel@edumazet-glaptop> References: <5057455A.7050108@googlemail.com> <50588371.40103@googlemail.com> <505885DC.1060006@googlemail.com> <1347979239.26523.267.camel@edumazet-glaptop> <5059E40C.4070607@googlemail.com> <505D5A18.2080507@googlemail.com> <50643DA1.7070306@googlemail.com> <1348748042.5093.1168.camel@edumazet-glaptop> <50649567.2010704@googlemail.com> <1348779826.5093.1750.camel@edumazet-glaptop> <1348780624.5093.1767.camel@edumazet-glaptop> <50656C4A.8090302@googlemail.com> <1348831592.5093.2251.camel@edumazet-glaptop> <5068648C.7050005@googlemail.com> <1349034330.12401.210.camel@edumazet-glaptop> <506955F3.8050304@googlemail.com> <1349082950.12401.669.camel@edumazet-glaptop> <5069B300.9080308@googlemail.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: David Miller , netdev@vger.kernel.org, gpiez@web.de To: Chris Clayton Return-path: Received: from mail-bk0-f46.google.com ([209.85.214.46]:42972 "EHLO mail-bk0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752391Ab2JAPbn (ORCPT ); Mon, 1 Oct 2012 11:31:43 -0400 Received: by bkcjk13 with SMTP id jk13so5036286bkc.19 for ; Mon, 01 Oct 2012 08:31:42 -0700 (PDT) In-Reply-To: <5069B300.9080308@googlemail.com> Sender: netdev-owner@vger.kernel.org List-ID: On Mon, 2012-10-01 at 16:13 +0100, Chris Clayton wrote: > > On 10/01/12 10:15, Eric Dumazet wrote: > > On Mon, 2012-10-01 at 09:36 +0100, Chris Clayton wrote: > >> > > > >> 0 ICMP messages received > >> 0 input ICMP message failed. > >> ICMP input histogram: > >> 0 ICMP messages sent > >> 0 ICMP messages failed > >> ICMP output histogram: > > > >> > >> After: > >> > >> $ netstat -s > >> Icmp: > >> 4 ICMP messages received > >> 4 input ICMP message failed. > >> ICMP input histogram: > >> echo replies: 4 > > > > So icmp replies come back and are delivered to host instead of being > > forwarded. > > > > I wonder if MASQUERADE broke... > > > > Could you send > > > > iptables -t -nat -nvL > > $ iptables -t -nat -nvL > iptables v1.4.15: can't initialize iptables table `-nat': Table does not > exist (do you need to insmod?) > Perhaps iptables or your kernel needs to be upgraded. > > > conntrack -L # while ping is running from guest > > $ conntrack -L > conntrack v1.2.2 (conntrack-tools): Operation failed: invalid parameters > Thats not expected, you described you used MASQUERADE target, so "iptables -t nat -nvL" should display something. > Forgive me for asking, but why is the problem not down to the change > that I identified by bisecting? The title of the patch is "ipv4: Cache > local output routes" and, although I'm a million miles from being an > expert here, to me it does make it look a good candidate. > http://marc.info/?l=linux-netdev&m=134797809611847&w=2 Because I cant reproduce your problem at all, using your setup.