From mboxrd@z Thu Jan 1 00:00:00 1970 From: Muriel Subject: Re: VM hangs after live migration Date: Wed, 12 Jan 2011 14:58:49 +0100 Message-ID: <4D2DB399.8000401@gmail.com> References: <4D2D7758.8050203@gmail.com> <1294827606.3831.6088.camel@zakaz.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15; format=flowed Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <1294827606.3831.6088.camel@zakaz.uk.xensource.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Ian Campbell Cc: xen-devel List-Id: xen-devel@lists.xenproject.org Il 12/01/2011 11:20, Ian Campbell ha scritto: > Hi Muriel, > > (I reordered the quoted material slightly) Hi Ian, thanks for your reply. > > On Wed, 2011-01-12 at 09:41 +0000, Muriel wrote: >> Hi all, >> i'm not sure that this is the right place to write but i hope you can >> help me. > > For XenServer specific issues you should generally either contact > XenServer support or make use of the appropriate XenServer forums rathe= r > than this list which is for the development of the upstream xen.org > version of Xen. Yes, i know, but the problem seems to be ambiguous, so i'm trying to ask=20 for an opinion to the different actors: - citrix for the environment - ubuntu for the guest - this list for the changes on the xen code I thought that this list is more correct than xen-user, but if you think=20 not so i can move the thread. If you think that both are wrong i'm in a=20 sea of trouble beacause: - citrix say "is a ubuntu problem" - it seems that for ubuntu "is a mainline kernel problem"=20 (https://bugs.launchpad.net/ubuntu/+source/linux/+bug/681083) - you say "is a citrix problem" end finally my boss say: "is a YOUR problem" :) > >> to the kernel 2.6.32.27 and now: >> - when i try to migrate the vm from a member of the pool to another >> everything works fine; >> - when i try to migrate the vm from a member of the pool to the master >> the vm hangs (exactly as the 2.6.32.16). > > If you see different behaviour based upon which hosts you are migrating > between then that, to me, suggests a XenServer issue in the first > instance. Maybe, but the problem appears with the 2.6.32.16 and this made me to=20 think that the bug has come with some modifications. I read the changelog and there are no significant changes from 15 to 16=20 (I'm wrong?). The only difference between the master and the members is that the=20 master has no other virtual machines. > >> I have many vm on Xenserver 5.5 and starting with kernel 2.6.32.15 i h= ad >> many problems with the live migration: >> - with 2.6.32.16 and higher the vm hang after migration (no console ou= tput) >> - 2.6.32.26 and higher have this kernel bug: >> kernel BUG at /build/buildd/linux-2.6.32/arch/x86/xen/spinlock.c:343! >> invalid opcode: 0000 [#1] SMP > > However if you have reason to believe that the issue is specific to the > guest kernel rather than the use of XenServer as a host then please > provide full details of the guest, i.e. the distribution in use, where > the kernel came from, the .config used to build it, the guest > configuration and full guest console logs etc. >> After some research on the mailing list i tried to apply this patch: >>=20 http://xen.1045712.n5.nabble.com/PATCH-xen-events-do-not-unmask-polled-ip= is-on-restore-td3241695.html > > There was a second patch at the end of that thread, which did you use? Sorry. guest: ubuntu 10.04 64-bit kernel: - ubuntu kernel 2.6.32.24 (based on 2.6.32.15 vanilla) -> works - ubuntu kernel 2.6.32.25 (based on 2.6.32.21 vanilla) -> works on=20 members, but not works members2master (guest hangs) - ubuntu kernel 2.6.32.26|27 (based on 2.6.32.24|26 vanilla) have the=20 bug above - vanilla kernel 2.6.32.27 have the bug above - vanilla kernel 2.6.32.27 with your 1=B0 patch works on members, but not= =20 works members2master (guest hangs) like 2.6.32.21 On what kernel i have to turn my attention to send log, etc (remember=20 that i not have console because the guest hangs)? > Have you tried the xen/stable-2.6.32.x branch from xen.git? No, but if can help i can try. My idea was to try to find the problem, send a patch to the Ubuntu team=20 and find the patch applied in the next ubuntu packages. > Thanks, > Ian. Thanks for your attention Ian. Muriel