Re: The two image formats called qcow

["Daniel P. Berrange" <berrange@redhat.com>]

On Wed, Mar 26, 2008 at 09:50:06AM +0100, Kevin Wolf wrote:
> Keir Fraser schrieb:
> > It's tricky where users' non-volatile storage is concerned though. Other
> > than that I would say the bug should be fixed immediately. Is there an easy
> > way to detect with reasonable reliability whether we have an old or new
> > image? Failing that we may have to provide a tool to convert old images to
> > new format.
> 
> Something like "that number looks too big, it be must little endian"
> could easily turn into "that harddisk looks big, let's break the image",
> I suspect.
> 
> However, I just noticed that the tapdisk qcow driver writes an extended
> Xen-specific header to the image file. This should be reliable enough to
> detect tapdisk images.
> 
> Is it an option to convert broken images to big endian when it is opened
> for the first time in ioemu? In this case, the fix for older versions
> could be in one place at least instead of being scattered over the whole
> file. Then you wouldn't be able to open such a file with tapdisk again,
> though.

I don't like the idea of  secretly migrating them to  the fixed disk format
without admin interaction / confirmation.

If we can detect the old style disk format, then perhaps we could put a
check into the hotplug scripts. That way, when the user tries to start
a guest with the old broken format, we could prevent the guest starting
and show them a error message. Then can then run the conversion tool
to fix the format & start the guest again. 

Dan
-- 
|: Red Hat, Engineering, Boston   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|