|
|
|
|
|
|
|
|
|
|
xen-users
Re: [Xen-users] Xen 3.3 hypervisor with stock dom0 kernel from Debian Et
2008/9/25 <xen-users@xxxxxxxxxxxxxxx>:
> Just out of curiosity: Are there any reasons why you want the xensource
> kernel when the debian kernel works OK?
Just a nagging feeling that the newest xensource kernels are more up
to date and suitable for Xen 3.3 - eg some of the new 3.3 (or even 3.2
and 3.1 for that matter) features might rely on code that isn't in the
Debian kernel. But nothing definite, as I don't know enough about the
potential changes.
> Super! How about using the debian lenny 2.6.26 domU kernel [1] on top of
> the debian etch 2.6.18 dom0 kernel [2] (with Xen 3.3)? I guess that would
> work?
Should do. The Lenny domU kernel would need to use the native
paravirtual ops stuff though - I haven't tried that yet myself. pvgrub
sounds like a good way to manage a Lenny domU kernel with an Etch dom0
kernel.
>
> And my last silly question: Are there any real benefits on using a 64 bit
> dom0 [3] when I am only serving 32 bit debian domU's?
Not sure - maybe if you want to run a 64bit domU in the future?. If
you can get all your memory recognised by the 32bit dom0, there might
be some advantages to using 32bit for both. Apparently you used to
need the same arch for both dom0 and domU to get certain
save/restore/migrate type features working, but I think 3.3 doesn't
have that limitation anymore. Also using the same arch for dom0 and
domU would make it easy for aptitude to update kernels for both (as
they'd share it), but with 3.3 pvgrub lets the domUs manage their own
kernels*, so that advantage goes away too (I think).
* yeah I know pygrub could do that too but it was a bit ugly, and
Debian never really embraced pygrub like RH/Suse did anyway.
But I'm a little fuzzy on all these details, so you'd want to get more
advice and/or test it yourself :)
--
Cheers
Anton
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
|
|
|
|