Agreed that we can’t yet everything but perhaps these may be sensible.

1. Latency regression because of a cold start related bug (might need at least 
two invokers and controllers)
2. For a single invoker (non black box), some expected l/sustained throughout 
range (rps) 
3. If we disable the black box cluster, might it be possible to catch some lb 
imbalance?


-r

> On Mar 19, 2018, at 7:52 AM, Christian Bickel <[email protected]> wrote:
> 
> As I said, we only have a very small travis machine.
> 
> But I think it's at least worth a try, if we get consistent performance 
> numbers in travis, to set a threshold and fail if it's below this threshold.
> 
> Greetings
> Christian
> 
> March 19, 2018 12:03 PM, "Rodric Rabbah" <[email protected]> wrote:
> 
>> Great - are you considering also a Travis CI matrix?
>> 
>> -r

Reply via email to