From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nikolay Aleksandrov Subject: Re: bonding (IEEE 802.3ad) not working with qemu/virtio Date: Thu, 28 Jan 2016 15:10:51 +0100 Message-ID: <56AA216B.3040105@cumulusnetworks.com> References: <20160128133350.GA2231@nanopsycho.orion> <56AA1D65.2000509@cumulusnetworks.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: netdev To: Jiri Pirko , =?UTF-8?Q?Bj=c3=b8rnar_Ness?= Return-path: Received: from mail-wm0-f51.google.com ([74.125.82.51]:38751 "EHLO mail-wm0-f51.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750794AbcA1OKx (ORCPT ); Thu, 28 Jan 2016 09:10:53 -0500 Received: by mail-wm0-f51.google.com with SMTP id p63so26281146wmp.1 for ; Thu, 28 Jan 2016 06:10:53 -0800 (PST) In-Reply-To: <56AA1D65.2000509@cumulusnetworks.com> Sender: netdev-owner@vger.kernel.org List-ID: On 01/28/2016 02:53 PM, Nikolay Aleksandrov wrote: > On 01/28/2016 02:33 PM, Jiri Pirko wrote: >> Mon, Jan 25, 2016 at 05:24:48PM CET, bjornar.ness@gmail.com wrote: >>> As subject says, 802.3ad bonding is not working with virtio network model. >>> >>> The only errors I see is: >>> >>> No 802.3ad response from the link partner for any adapters in the bond. >>> >>> Dumping the network traffic shows that no LACP packets are sent from the >>> host running with virtio driver, changing to for example e1000 solves >>> this problem >>> with no configuration changes. >>> >>> Is this a known problem? > Can you show your bond's /proc/net/bonding/bond ? And also in order to > better see what's going on I'd suggest enabling the pr_debug() calls in the 3ad > code: > echo 'file bond_3ad.c +p' > /sys/kernel/debug/dynamic_debug/control > (assuming you have debugfs mounted at /sys/kernel/debug) > Then you can follow the logs to see what's going on. > I can clearly see LACP packets sent over virtio net devices: > 14:53:05.323490 52:54:00:51:25:3c > 01:80:c2:00:00:02, ethertype Slow Protocols (0x8809), length 124: LACPv1, length 110 Nevermind my message, Jiri was right. My setup wasn't correct. I have a virtio_net get_settings hack that enables me to test 802.3ad over virtio_net locally. Sorry about the noise. > >> >> I believe the problem is virtio_net for obvious reasons does not report >> speed and duplex. Bonding 3ad mode makes that unconfortable :) [snip] Right, of course. My setup has some patches that add get_settings to virtio_net to make it play nice for testing 803.2ad. >> >> Use team ;) > :-)