Re: [Xen-users] HVM failing. Tips please?

[Aleix Dorca <adorca@xxxxxx>]

Hi Ken!

thanks for answering so fast. My Dom0 uses all CPUs available (8 in  
this case, due to a bug that doesn't allow VMs to shutdown, reboot...  
if I set this to 1) so it's hard for me to separate DomU from Dom0 in  
CPU usage, although it's a good idea.
Thanks,

Aleix.

El 25/03/2009, a las 16:16, Ken Cobler escribió:

> Aleix Dorca wrote:
> > Hi people,
> >
> > I have several Xen Dom0's working. Those that have PV domU's work  
> > awesomely and without any problems. The ones that have HVM domU's  
> > fail once every two weeks (more or less). Here's the setup:
> > Dell PowerEdge 1950 III
> > Ubuntu 8.04.1 (64bits) with Xen 3.3.0 compiled over the linux-image- 
> > xen from Ubuntu.
> > HVM domU's are Windows 2003 Server Enterprise
> > I've set the clocksource to "jiffies" to avoid high load fails  
> > (this reduced the failing A LOT).
> > All disks are accessed via iscsi.
> >
> > As said every now and then Dom0 hangs completely. No net access  
> > whatsoever... all gone. I have to manually stop the machine and  
> > restart it.
> I have Suse 10.0 (Xen 3.1) running with Windows 2003 and Windows XP  
> VMs.  Both Windows 2003 and Windows XP are running from boot.
> I have found that I pin a VM to a vCPU.  In HVM, all disk access (if  
> not using GPLV driver) uses dom0 to complete the request.  By making  
> certain that dom0 has 1 vCPU to itself always, limits the chance of  
> getting into an I/O bottleneck.
> By default, all domU's and dom0 are allowed to run on any vCPU  
> available.
> Ken Cobler
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users