From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: Re: Jeremy's GIT-tree and network problems Date: Tue, 30 Mar 2010 10:30:06 -0400 Message-ID: <20100330143006.GD10728@phenom.dumpdata.com> References: <20100329T1647.GA.2c139.stse@fsing.rootsland.net> <20100329181815.GF17315@alpha.trueline.nl> <20100330T1423.GA.57721.stse@fsing.rootsland.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: <20100330T1423.GA.57721.stse@fsing.rootsland.net> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: xen-devel@lists.xensource.com Cc: stse+xen@fsing.rootsland.net, Adnan Misherfi List-Id: xen-devel@lists.xenproject.org On Tue, Mar 30, 2010 at 02:26:00PM +0200, Stephan Seitz wrote: > On Mon, Mar 29, 2010 at 08:18:15PM +0200, S.H. Verbrugge wrote: >> Could you check your dmesg (dom0) to see if you have similar messages? >> http://lists.xensource.com/archives/html/xen-devel/2010-03/msg00966.ht= ml > > You mean =E2=80=9EAttempting to checksum a non-TCP/UDP packet, dropping= a=20 > protocol 1 packet=E2=80=9D? Yes, I see one shortly before the NFSv4 ser= ver is =20 > started. But it is only one line and doesn=E2=80=99t show again if I st= art my =20 > connection tests. So here is an interesting observation (And I am late in this thread so if I am going off on a tangent, please correct me): We compiled both in DomU and Dom0 the kernel to not have any traces of iptables nor igbtables. The performance of scp-ing a tarball from Dom0 to DomU went from 0.5MB to 17MB/s (which is about the disk speed). I don't know if you feel inclined to actually try such a drastic option to see if that is your culprit or not, but it is worth keeping in the back of your head.