On Thu, Jun 29, 2006 at 05:14:11PM -0600, Chris Vaughan wrote:
> So what's the difference between a "Live Migration" & "Save" then? Does the
> live migration keep the image open and the save does not? To me they seam
> to be 2 peas in a pod but I guess I'm mistaken.
Live migration ensures that the filesystem is passed as is from one
state to the other. Restoring a checkpointed system would mean that
the system would suddenly warp back in time wrt to the filesystem, or
what is effectively the same from the system's POV the filesystem
suddenly warped ahead in time.
For example: Perhaps your crash between the checkpointing and the
resuming was caused by rm -fr / and no checkpointing could undo that :)
So checkpointing in xen would only work hand in hand with block device
snapshoting.
> Thanks
>
>
> On 6/29/06, Daniel Veillard <veillard@xxxxxxxxxx> wrote:
> >
> >On Thu, Jun 29, 2006 at 04:56:21PM -0600, Chris Vaughan wrote:
> >> Hi,
> >>
> >> Is there any way you can save a xen domain without suspending it or
> >stopping
> >> it? I want to set up something so if my xen session crashes I can fire
> >up
> >> another xen session from the save file that I created 5 minutes before
> >the
> >> crash and hopefully minimize downtime.
> >
> > Doesn't work in general because you would have to save the state of the
> >filesystems too. Xen operations don't allow a save and continue for that
> >reason I was told.
> >
> >Daniel
> >
>
>
>
--
Axel.Thimm at ATrpms.net
pgplLkW7kl7Jg.pgp
Description: PGP signature
|
