Return-path: Received: from mail-iy0-f174.google.com ([209.85.210.174]:65310 "EHLO mail-iy0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755156Ab1DURYo convert rfc822-to-8bit (ORCPT ); Thu, 21 Apr 2011 13:24:44 -0400 Received: by iyb14 with SMTP id 14so1563542iyb.19 for ; Thu, 21 Apr 2011 10:24:43 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: References: From: Javier Cardona Date: Thu, 21 Apr 2011 10:24:23 -0700 Message-ID: Subject: Re: Mesh broken To: Qasim Javed Cc: linux-wireless , ath9k-devel@lists.ath9k.org, devel@lists.open80211s.org, "Luis R. Rodriguez" Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Mon, Apr 18, 2011 at 2:18 PM, Qasim Javed wrote: > Hi all, > > We have a 6 node mesh setup, with one node acting as a mesh portal. > All the nodes are in direct range of each other, so every node has 5 > peer links. The node that acts as a mesh portal has another laptop > connected to it via an Ethernet cable which runs an iperf TCP server. > Every mesh node runs an iperf TCP client where the transmit data rate > is limited to about 500Kbps. > > When we run the iperf client on all the mesh nodes while using > compat-wireless from 18th Jan, 2011, things seem to work fine. There > is rarely a connection loss. However, the peak data rate at the laptop > which is connected to the mesh portal (via Ethernet) goes much higher > than the expected 3Mbps (6*500Kbps). > > When we use compat-wireless from 28th March, 2011, and repeat the > above experiment again, the results we get are quite disappointing. > The TCP connections from all 6 nodes constantly break. Also, if only > one mesh node is communicating with the laptop connected to the mesh > portal, the connection does not break. If using the same > compat-wireless, we force the max plinks to 1 (one) at each node, the > TCP connections do not break. Everything works fine. > > If we use compat-wireless from 31st march, 2011, the mesh nodes are > not even able to establish plinks with each other. > > "iw dev mesh station dump" > > returns nothing. > > Switching to compat-wireless from 14th April, 2011 results in plinks > alternating between LISTEN and OPN_SNT states. The displayed beacon > signal strength (in station dump) is very low (close to -80dbm). > > Infrastructure mode works fine. > > Can anyone please elaborate on what could be happening here? I have no clue, but FWIW we tested a recent version of wireless-testing (1d4af019d920c838e4de81afa0b5d67879090fd5) on ath9k hardware (AR5418) and worked well?. Javier [1] We had to apply a fix to that particular card that (http://o11s.org/trac/wiki/ath9kFilteringIssueFix) but that does not seem to apply to your hardware as you were running in mesh mode before.