On Wed, May 11, 2016 at 4:19 PM, David Caro <dcaro@redhat.com> wrote:

You are not having into account yet that we are not using most of the hardware
we have yet, that will allow us to have more than twice the amount of slaves we
have now too (given that we solve any other bottleneck, like nfs storage)

I agree. I think best thing to begin with would be to add as many slaves

as we can and remove all restrictions from the jobs and monitor the queue.

Last time we tried using it, it ended messing up all the executions, mixing
slaves and creating a lot of failures, if you try it out, be very vigilant

good to know