Assuming the applications can tolerate EAGAIN errors and the guest
VM’s front end keep returning EAGAIN for the period of dom0 going down
and coming back up, the drivers don’t need to move to Xen/other domain. Also
if we have a driver domain which I understand is restartable, how is the
pending IO handled ?.
Thanks,
-SK.
From: Charlie Smurthwaite
[mailto:charlie@xxxxxxxxxxxxxx]
Sent: Friday, April 30, 2010 9:55 AM
To: Subbiah Kandasamy (skandasa)
Cc: Pasi Kärkkäinen; xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-devel] Dom0 restartability
Am I correct in believing that
the majority of IO for all domains passes through the DOM-0 kernel? I would
think that a huge number of IO drivers would need to move into Xen in order to
allow this without severe IO disruption. Are there methods of disk / network IO
that do not rely on DOM-0?
Charlie
On 30 April 2010 17:19, Subbiah Kandasamy (skandasa) <skandasa@xxxxxxxxx> wrote:
Thanks Pasi, I am sure there are other things as well. I am
looking for any pointers listing the work done in that area.
Thanks,
-SK.
-----Original Message-----
From: Pasi Kärkkäinen [mailto:pasik@xxxxxx]
Sent: Thursday, April 29, 2010 5:34 PM
To: Subbiah Kandasamy (skandasa)
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-devel] Dom0 restartability
On Thu, Apr 29, 2010 at 04:05:13PM -0700, Subbiah Kandasamy (skandasa) wrote:
> In today's Xen summit I heard there is some work done on the
> restartability of dom0. Any pointers to what has been done
and what is
> pending and the patches ?.
>
Restarting dom0 requires for example moving xenstored out of dom0..
so you don't lose the database when you restart dom0.
-- Pasi
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
