Hi!

On Mon, Apr 8, 2013 at 11:25 AM, Guido Trotter <[email protected]> wrote:

> qq: ensuring consistency is hard, would it be bad to allow for it to
> be the option (and the lvm volume, etc) even if they are disabled, and
> in that case they should just be unused (and maybe not displayed by
> cluster info)?
>

I was wondering about that, too. In this patch, I restored the behavior
that was present before with the two options. However, I also find it
reasonable to have a vgname and a usermode helper specified even if those
disk templates are not enabled to prepare for when they get enabled. I
would rather widen that once I have refactored more of the lvm-based design
of ganeti, because right now vgname is the way to dis/enable drbd and I
want to do one step at a time. Ok?

Cheers,
Helga


>
> Thanks,
> Guido
>
> On Mon, Apr 8, 2013 at 9:58 AM, Michele Tartara <[email protected]>
> wrote:
> > On Wed, Apr 3, 2013 at 2:57 PM, Helga Velroyen <[email protected]>
> wrote:
> >>
> >> This assures that the consistency of the options
> '--drbd-usermode-helper'
> >> and '--enabled-disk-templates' of 'gnt-cluster modify' is checked
> >> correctly.
> >> Also, this makes the drbd usermode helper configurable in the QA config.
> >>
> >> Signed-off-by: Helga Velroyen <[email protected]>
> >> ---
> >>  qa/qa-sample.json |  2 ++
> >>  qa/qa_cluster.py  | 21 +++++++++++++++++++++
> >>  2 files changed, 23 insertions(+)
> >
> >
> > LGTM, thanks.
> >
> > Michele
>
>
>
> --
> Guido Trotter
> Ganeti engineering
> Google Germany
>

Reply via email to