On Tue, Dec 25, 2012 at 03:26:55PM -0800, Vincent Janelle wrote:
Is there a way of determining spec testing reports of manifests? I'm
adding tests to an existing deployment and it'd be useful to find out what
code isn't being tested.
I am not aware of such a tool. Also, I do not think that this is easily
possible. Testing (both with rspec and cucumber) puppet means checking
the compiled catalog. A mapping back to the declarative manifests which
are also split into multiple files trhough modules and assembled back
together by types and classes is difficult.
Personally, I set up a guard (file-system watcher) that executes the
tests when I change something. For testing existing declarations, I have
the following process:
1. write the test that resembles the current state
2. if the new test is green, I check by commenting out the puppet code.
3. repeat step 1 and 2 until the test fails without the puppet code
4. commit the new test into version control, have the production active
again and run all tests to see if the test also works in the complete
5. after commiting to version control, I use the opportunity to clean up
the code a little, maybe document better or something like that, but
that's the developer in me. :)