From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Lalancette Subject: Re: Migrating Vm's from one machine to another...and back Date: Mon, 11 Aug 2008 19:01:26 +0200 Message-ID: <48A07066.9000108@redhat.com> References: <48A05869.60706@bull.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org, Laurent Vivier To: Jean-Pierre Dion Return-path: Received: from mx1.redhat.com ([66.187.233.31]:47291 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756239AbYHKRDy (ORCPT ); Mon, 11 Aug 2008 13:03:54 -0400 In-Reply-To: <48A05869.60706@bull.net> Sender: kvm-owner@vger.kernel.org List-ID: Jean-Pierre Dion wrote: > Hi, > > I made some VM migrations between several machines > (Intel-Intel and Intel-AMD) and when one VM has been > migrated onto a second machine, I cannot put it > back on the first one. I talked with Laurent and it > seems that there is no way (or no obvious one) > to do that. > > This could be useful when doing some tasks (load > balancing...). > > Is it an item that is interesting enough to be done ? > And does anyone have some idea on how to do that ? Hm, this seems to work for me, in combination with my recently posted "Implement tcp nowait option" patch. Basically you start the guest on machine A like normal. Then you start the "container" on machine B with the -incoming flags. Now you migrate the guest over to machine B (I assume you've gotten this far). Now, you kill the container on machine A (since the running guest is on B now), and restart it with -incoming. Now you migrate back from B to A, and it all works out. You can actually probably get rid of one of these steps by starting the machine on machine A in the first place with -S -incoming tcp://0:4444,nowait, and then running "cont" in the monitor. In this case, the machine is already running, and *also* listening for incoming migrate requests. Now you do exactly the same on machine B, and you should be able to migrate back and forth between them at will. Chris Lalancette