Also check out the ‘Background’ feature if you want to avoid duplicating data, this wold have helped in our devopsdays session.

With pre-discovery this finishes in:

1 scenario (1 passed)
9 steps (9 passed)
0m8.771s

Cucumber though might not really be the best fit, I don’t like the data redundancy between Scenarios need to spend more time with its docs but another testing framework might be needed. Also of course cucumber scenarios are stand-alone, I need to either build out a huge nasty scenario or find a way to have the previous one bail out the next. Lots of time wit the docs needed.

The real win for me is that I have improved all the various mcollective bits to return true/false or to be able to request their status. This is critical for building actual deploy systems with it. When cucumber says blocking the ip worked I now know it worked without a doubt on every webserver mcollective found.

I think using test frameworks for this is probably a sane idea, I just need the right one

The above scenarios:

3 scenarios (3 passed)
8 steps (8 passed)
0m14.519s

I can speed that up a lot by removing some redundant discoveries. But for that I need a global counter of how many hosts are found in the initial discovery, hard with cucumber.

Excellent work! I’m really digging the mcollective stuff by the way.