maas-devel team mailing list archive

[Diogo Matsubara <diogo.matsubara@xxxxxxxxxxxxx>]

On Wed, Apr 9, 2014 at 9:43 AM, Mark Shuttleworth <mark@xxxxxxxxxx> wrote:

>  On 09/04/14 13:29, Diogo Matsubara wrote:
>
>  On Wed, Apr 9, 2014 at 6:41 AM, Mark Shuttleworth <mark@xxxxxxxxxx>wrote:
>
>> On 09/04/14 04:35, Julian Edwards wrote:
>> > 5. Verify that the lab tests continue to pass when they run with your
>> > change (they use the daily build PPA package).  Subscribe to the
>> > maas-builds mailing list to get the daily build reports.
>>
>>  Do these tests include an automated build of an OpenStack, on all
>> current LTS releases?
>>
>
>  No, they include a deployment of the mediawiki charm on top of the MAAS
> cluster.
>
>
> Thanks Diogo. Is the constraint the amount of hardware we have? In which
> case, can we do this weekly on the gMAAS?
>

Last I read[1], we'd need a lot of hardware, yes, even if we don't do the
full HA stuff.
That lab has 12 machines (one is the main entry point to the lab and
another one doesn't have a disk, so essentially 10 nodes) and now that juju
supports LXC, I think it's possible to deploy the charms in less nodes, so
it's a matter of trying and see how it goes and then update the tests to
automate. That said the lab is also used by the Red Squad, some members of
the HWE team and I to try/test things out with MAAS, so it's a scarce
resource that we try to manage usage among ourselves and the automation.

You suggest using gMAAS to do this automated or manually on a weekly basis
and then report back the findings? I guess gMAAS suffers from the same
issue above, people are competing all the time to use the resource...
In any case, I'll discuss this with Pat, figure out a plan and let you know
how we're going to tackle this.

[1] https://wiki.ubuntu.com/ServerTeam/OpenStackHA


-- 
Diogo M. Matsubara