From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nivedita Singhvi Subject: Re: help with horrible network failures Date: Wed, 02 Mar 2005 10:01:14 -0800 Message-ID: <4225FF6A.8080104@us.ibm.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit In-Reply-To: Sender: xen-devel-admin@lists.sourceforge.net Errors-To: xen-devel-admin@lists.sourceforge.net List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , List-Archive: To: Ian Pratt Cc: Rob Gardner , xen-devel , ian.pratt@cl.cam.ac.uk List-Id: xen-devel@lists.xenproject.org Ian Pratt wrote: >>I've got two machines running identical versions of xen & >>linux. One of >>them constantly has problems under high network loads >>(70-100% httperf >>loads). This is using xen-unstable checked out on or about >>Feb 13. Any >>clues? See below for details. > > > One machine exhibits the bug, the other doesn't? How similar is the h/w? I believe this is a known unresolved bug in mainline. It's possible Rob is provoking it by using NAPI - and it is load-related, so that might explain why one box is seeing it and one isn't. Essentially, data is getting reordered where it really shouldn't, due to a race that isn't clear. It would be very helpful to track this down finally - if it hasn't already.. > Can you reproduce with a single high-rate TCP stream? Do you have any of > the iptables/netfilter connection tracking stuff in your kernel? > Can you reproduce with an older version of Xen? I suspect that it is independent of Xen - unless some versions of Xen slow down traffic enough to change the window.. thanks, Nivedita ------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click