[Gllug] Oddity with virt-install

[Daniel P. Berrange]

On Tue, Sep 23, 2008 at 06:42:32AM +0100, Stephen Nelson-Smith wrote:
> Hi there,
> 
> > Your installation args appear to be a reasonable
> 
> Aha - but on a closer look they aren't!  Why would I need to specify
> -s 4 if I've already passed it a logical volume?  Well, because the
> logical volume doesn't exist - it's hb-test1 not ha-test1!
> 
> Having changed that, it all ran smoothly.  Some kind of message would
> have been helpful rather than silently failing.  I'm not sure where
> that knowledge should be though, does virt-install check if the
> logical volume exists?  I appreciate it asks for the size, on the
> command line, if you don't provide an image or logical volume of a
> given size, but if the logical volume suggested doesn't exist, it's
> not going to get lvm to make one, so we're onto a loser straight away
> aren't we?

Ah so the reason it wouldn't give you a error is because you had both a
non-existant disk path, and a size - so it assumed you wnated to create
a file - if you'd left off -s 4  it would have told you the file didn't
exist.

BTW, one tip is to use  --debug if things go wrong - it'll dump out 
lots of info - at the very least you can attach it all to a BZ ticket
which will help us diagnose problems.

> > I see you've got a i686 install tree for the guest - if you are trying to
> > run this on a x86_64 host (ie a so called 32-on-64 install), then make
> > sure you are using at least CentOS 5.2 - paravirt 32-on-64 is very flaky
> > in Xen and RHEL-5.2 is first time it was reasonably stable.
> 
> Yes, I'm using RHEL 5-2 (and CentOS) this way.  Given that I don't do
> any migration stuff I find it perfectly stable.  Wonder if it'll leave
> technology preview soon?

In RHEL-5.1 it was classed tech preview because it was known to randomly
crash with doing paravirt 32-on-64.

In RHEL-5.2 we fixed the random crashes, but it remains tech preview
because it is lacking features - save/restore/migration - if you're not
using those features then it ought to do the job for you reasonably well.

I don't expect that to change in RHEL-5.3, potentially not even in a future
update - its very tricky to cherry-pick the neccessary fixes without 
causing regressions in other areas.

Daniel
-- 
|: http://berrange.com/     -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://freshmeat.net/~danielpb/    -o-   http://gtk-vnc.sourceforge.net :|
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug