Open Stack

On Wed, Jun 12, 2013 at 6:32 PM, Nikola Đipanov <ndipanov at redhat.com> wrote:
> On 12/06/13 00:11, Michael Still wrote:

>> Perhaps tweaking this migration is ok as long as its functionally
>> identical -- early deployers just suck up some pain, and later
>> deployments run a bit faster. Is that possible?
>
> I think this is the best option - I will not change the migration
> functionally - just tweak it's performance (hopefully). I will be
> posting a patch later today (in CET tz) so you can run it on your
> dataset and see if there are any significant improvements.
>
> I would definitely prefer this over reverting/removing the migration,
> provided we can make good (or any) performance gains.

Agreed. For reference, any patchset proposed with a change to a file
in nova/db/sqlalchemy/migrate_repo/versions will get run through my
tester automatically. Results are posted to
http://openstack.stillhq.com/ci/

The eventual plan is to have this code comment on reviews like
smokestack or jenkins, but the code is very new so I want to make sure
its right first.

However, restating a migration in a patchset probably wont work with
my test script, as master is applied first. That's something I should
probably fix in my tester. I'll take a look at that soon.

Cheers,
Michael