[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] old issue after 1024 live migrations seems to still exist.



On Wed, Jul 21, 2010 at 04:45:08PM +0000, Vern Burke wrote:
> I've been running my own load balancer right along on XCP and never have seen 
> this.
> 

Yeah, people on xen-devel said this is not expected behaviour,
and they haven't seen it.

So it's probably something specific to this exact OracleVM version or so.
We'll see when people investigate it further.

-- Pasi


> Vern
> 
> Vern Burke, SwiftWater Telecom, http://www.swiftwatertel.com
> 
> -----Original Message-----
> From: Boris Quiroz <bquiroz.work@xxxxxxxxx>
> Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
> Date: Wed, 21 Jul 2010 12:25:59 
> To: Florian Heigl<florian.heigl@xxxxxxxxx>
> Cc: Xen Users<xen-users@xxxxxxxxxxxxxxxxxxx>
> Subject: Re: [Xen-users] old issue after 1024 live migrations seems to still 
>       exist.
> 
> 2010/7/21 Florian Heigl <florian.heigl@xxxxxxxxx>:
> > Hi,
> >
> > last month I did some checkig of old Xen issues that I remember and
> > found this one to still exist - if you do a high amount of live
> > migrations at some point the xen daemon chokes and dies.
> > The issue was reported by someone on the list like 4-5 years ago, but
> > it seems it hasn't been fixed (not sure if anyone even replied back
> > then)
> > The Xen version I used to test as 3.4.0 from Oracle VM 2.2
> >
> > Basically You just ping-pong one domU and somewhere after 900
> > migrations you first see it drop the ball a few times (vm needs to be
> > restarted) and then about 100 times later one one of the hosts the xen
> > daemon will crash, restart and not be able to boot vm's any more.
> >
> > (I waited a while to post this, but about time now I get it done)
> > I'm building some power management magic witrh loadbalancing so that
> > idle servers can automatically shutdown and startup, and cpu intensive
> > vm's can be distributed evenly.That this bug still exists is a
> > nightmare: 1024 migrations sounds a lot, but with 128 VMs on a host it
> > just equals just 4 migrations per VM, right? Without the loadbalancing
> > bit this wouldn't have to happen very often, but I think it's a key
> > feature.
> > If the RDMA live migration ever comes around, there'd be nothing against 
> > it...
> >
> > I've also prepared a clumsy script for the test, which can be found here:
> >
> > http://wartungsfenster.pastebin.org/410803
> >
> > I can open a bug report but i think it'd be best if someone re-test on
> > Xen4 first.
> >
> > Regards,
> > Florian
> >
> > p.s.:
> > why is live migration so slow (2-3 seconds)  - without sdp i had 2-3
> > gbit of bandwidth, the vm was 64MB size  (that means 1/6 second of
> > transfer for the main bulk) and idle without networking!
> > is it just the gratious arp?
> >
> > --
> > 'Sie brauchen sich um Ihre Zukunft keine Gedanken zu machen'
> >
> >_______________________________________________
> > Xen-users mailing list
> > Xen-users@xxxxxxxxxxxxxxxxxxx
> > http://lists.xensource.com/xen-users
> >
> 
> Creepy..
> If this bug is still unsolved, could be XCP buggy too? Can anybody
> test it? At work I don't have any testing environment =(
> 
> Cheers.
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users

> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.