Re: dom0 from ubuntu to NetBSD?

[Matthias Petermann <mp%petermann-it.de@localhost>]

Hello Jukka and Pierre-Phillipp,

Please excuse if this email does not appear in the original thread. I am 
new to this list and am trying to reply to the email from the archive.

> Theoretically you would only loose sparse file capability.
> And if it was a XEN cluster then you would also need to consider some
> network storage specifics.

Pierre-Phillip's advice is valuable and correct. I would like to add 
that sparse files and Xen also work on NetBSD.


jupiter$ dd if=/dev/zero of=nbsd10.img bs=1 count=1 seek=128g
1+0 records in
1+0 records out
1 bytes transferred in 0.001 secs (1000 bytes/sec)

...

jupiter$ ls -lahs nbsd10.img
 25G -rw-r--r--  1 root  users  128G Jul 26 18:49 nbsd10.img

It is only important to note that the cp command fills in the gaps in 
the sparse file and blows it up to the full size. To copy the file, pax 
or tar are more suitable as they understand the characteristics of a 
sparse file.

What is currently not working - and probably that was what was meant - 
copy on write a la qcow2. This requires the blktap driver, which is not 
available in NetBSD.

I use Xen here in mixed operation with some file-based storages 
according to the "sparse file" approach indicated above. And I also use 
LVM volumes. In my experience, LVM is faster, but more is noticeable - 
especially if you still use mechanical hard drives - if the storages are 
spread over several independent drives.

Kind regards
Matthias