hi folks, is there a straightforward or well-documented way to migrate my physical sxce snv_115 (x64) system into an xvm ? searching for "p2v" in an opensolaris context seems to pick up a few hits on zones, but nothing obvious relating to xvm on opensolaris for what it''s worth the host system is opensolaris (2010.02 snv_126), but i''m hoping that''s not very significant (?) moving physical systems into virtualised infrastructure seems to be pretty popular these days - i''m surprised there isn''t more documentation on it - or have i missed something obvious? p -- This message posted from opensolaris.org
Peter Lees wrote:> hi folks, > > is there a straightforward or well-documented way to> migrate my physical sxce snv_115 (x64) system into an xvm ? It''s not that hard to do.. But it is not well documented.> searching for "p2v" in an opensolaris context seems to> pick up a few hits on zones, but nothing obvious relating > to xvm on opensolaris> > for what it''s worth the host system is opensolaris> (2010.02 snv_126), but i''m hoping that''s not very > significant (?) unfortunately, a zfs root makes this a bit harder to do... Without a zfs root, you had to update bootpath in bootenv.rc, update vfstab, and update /etc/hostname.* and/or /etc/dhcp.* With a zfs root, you need to attach the disk to another PV guest (non zfs root) using the device it will use when you start the disk (i.e. xvda). Then import the rpool, update /etc/hostname.* and/or /etc/dhcp.*. The disk should be ready to boot as a PV guest at this point. zfs encodes the bootpath in the zpool meta data. When you import a pool, it updates the meta data.> > moving physical systems into virtualised infrastructure> seems to be pretty popular these days - i''m surprised > there isn''t more documentation on it - or have i missed > something obvious? no, nothing available today.. It is being worked on... MRJ
On 11/17/2009 6:48 PM, Peter Lees wrote:> Mark Johnson wrote: >> >> >> With a zfs root, you need to attach the disk to another >> PV guest (non zfs root) using the device it will use when >> you start the disk (i.e. xvda). Then import the rpool, >> update /etc/hostname.* and/or /etc/dhcp.*. The disk >> should be ready to boot as a PV guest at this point. >> zfs encodes the bootpath in the zpool meta data. When >> you import a pool, it updates the meta data. > > > thanks mark, > > so you''re saying i need an existing non-zfs-root paravirtualised guest > to start?yes... There may be a a way getting around the zfs limitation of having two pools named rpool.. Not sure... For me, it easier to use a guest without a zfs root.> the question i have is: how to actually move the zfs pool from the > original into a volume on the xvm host? afaik i can''t do a zfs send from > a filesystem to a volume: > > # zfs send -R rpool@today \ > | ssh xvm-host zfs recv -u storage/guests/newvm/disk1 > > ... or can i ?It would be easier to do a dd :-) But it should be possible with some extra work to zfs send/recv.. I''ll have to give that one a try. MRJ> > p