vmbuilder strange outcome

[Serge E. Hallyn]

Quoting Dan Sheffner (dsheffner at gmail.com):
> actually both files for vm00 are gone.  I also tried this again without
> using tmpfs and got the same results.  It seems like creating the second vm
> is deleting the files for the first vm.  Is this even possible?
> 
> 
> On Fri, Jul 30, 2010 at 9:21 AM, Serge E. Hallyn <serge.hallyn at canonical.com
> > wrote:
> 
> > Quoting Dan Sheffner (dsheffner at gmail.com):
> > > I then do a reboot on the actual server running both virtual machines.
> >  When
> > > the server comes backup vm01 is running and vm00 is not.  Then when I try
> > to
> > > start vm00 I get:
> > >
> > >  Id Name                 State
> > > ----------------------------------
> > >   1 vm01                 running
> > >   - vm00                 shut off
> > >
> > > virsh # start vm00
> > > error: Failed to start domain vm00
> > > error: monitor socket did not show up.: Connection refused
> >
> > Interesting - is there any helpful info under /var/log/libvirt?
> >
> > > Maybe I'm missing something trivial.  Please let me know.

Doh, yes, something trivial indeed.

At least here on lucid, the vm gets placed inside $CWD/ubuntu.kvm.
That is, the second vmbuilder argument + '.' + the first argument.
You used the same for both When you run the command a second time,
it deletes that directory and creates a new one!

So, my recommendation:

mkdir vm1; cd vm1; (first_vmbuilder_command); cd ..
mkdir vm2; cd vm2; (second_vmbuilder_command); cd ..

Doh.

-serge