commenting a section breaks salt without a proper error message

[FlorianLudwig]

Adding:

  'something':
# nothing

to your top.sls will break salt. Running salt commands will give you no output (not even warnings or errors!). Running the minion with -l debug will bring up:

Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/salt/minion.py", line 635, in _thread_return
    ret['return'] = func(*args, **kwargs)
  File "/usr/lib/python2.7/site-packages/salt/modules/state.py", line 201, in highstate
    cache_name=kwargs.get('cache_name', 'highstate')
  File "/usr/lib/python2.7/site-packages/salt/state.py", line 2173, in call_highstate
    err += self.verify_tops(top)
  File "/usr/lib/python2.7/site-packages/salt/state.py", line 1789, in verify_tops
    for slsmod in slsmods:
TypeError: 'NoneType' object is not iterable

[UtahDave]

I think this may have been fixed by this commit: #6525

@FlorianLudwig can you test against git develop?

[basepi]

I think this is just the fault of an improper top.sls. Can you paste your whole top.sls?

I'm seeing a dictionary key that is set to None. Then when we iterate through, we expect a list there, and try to iterate over it.

[FlorianLudwig]

@basepi you are correct it is "just" a faulty top.sls. As said you can produce it with any top.sls by addign those to lines:

  'something':
# nothing

In my humble opinion this should not break salt without any error message.

@UtahDave sems still broken to me

[basepi]

But.....it did have an error message, right? You would just like an error message that was more informative?

(Not trying to be difficult, just trying to ascertain exactly what you're looking for)

[FlorianLudwig]

@basepi no, i mean there is no error message. There is a traceback on the minion - but running salt highstate does return exactly nothing.

[basepi]

Oh, well, that changes everything. I assumed we were returning that traceback or at least reporting that state compilation failed on the CLI.

Should be an easy fix. Thanks, and sorry I misunderstood.

[FlorianLudwig]

@basepi Maybe that is a bug worth on its own? So that tracebacks get send back to the master in case something went wrong.

[basepi]

Well the key is that we don't want to do a global try-except. We have a list of errors which we send back to the master (which is what I used to fix this issue). We just need to be vigilant for further holes in our error-checking.