Open Stack

On 15 February 2014 12:21, James E. Blair <jeblair at openstack.org> wrote:

> You won't end up with -1's everywhere, you'll end up with jobs stuck in
> the queue indefinitely, as we saw when the tripleo cloud failed
> recently.  What's worse is that now that positive check results are
> required for enqueuing into the gate, you will also not be able to merge
> anything.

Ok. So the cost of voting [just in tripleo] would be that a) [tripleo]
infrastructure failures and b) breakage from other projects - both
things that can cause checks to fail, would stall all tripleo landings
until rectified, or until voting is turned off via a change to config
which makes this infra's problem.

Hmm - so from a tripleo perspective, I think we're ok with this -
having a clear indication that 'this is ok' is probably more important
to us right now than the more opaque thing we have now where we have
to expand every jenkins comment to be sure.

But- will infra be ok, if we end up having a firedrill 'please make
this nonvoting' change to propose?

-Rob

-- 
Robert Collins <rbtcollins at hp.com>
Distinguished Technologist
HP Converged Cloud