Puppet Community

I 'borrowed' a function from some deprecated mode
```Puppet::Parser::Functions::newfunction(:is_classified_with, :arity =&gt; 1,
                                       :type =&gt; :rvalue) do |(str)|
  compiler.node.classes.keys.include?(str)
end```

That likely shows which classes were added to the node definition, usually by the ENC.

Which is related to which Classifier Groups the node was matched to, but not quite the same.

PE uses some secret sauce in classification, so `defined`  function never worked to detect those classes

Gang is there a way in Puppet to define an external fact combined with a dependency?  "Dont define this fact until, this other thing is already completed on the system?"

```set -e
/opt/other/thing do stuff
mkdir -p /etc/puppetlabs/facter/facts.d
echo "other_things_done=true" &gt; /etc/puppetlabs/facter/facts.d/other.txt```


A fact, is by definition, information about a current state.  So, if the thing isn't completed, the fact should return something that identifies that, I would think.  Can you give us a more complete example?

As gross as this looks ... I have made this an external fact...  but it requires python3 to be installed to do the thing.

``` Facter.add('python3_installpath') do
    confine kernel: 'Linux'
      setcode 'python3 -c "import sys; print(sys.path[4])" | sed "s/dist-packages//g"'
    end```

So my head says .. find away to make this fact only define after I install python

What are you trying to accomplish?  Wouldn't your other install thing just require python?

Hmmm looking into on how to confine... to python...

I need the fact so I can manipulate files within the python3 path which could change with python updates...  I know its gross...

<https://github.com/voxpupuli/puppet-python/blob/master/lib/facter/python_version.rb>

so you can do
```confine { Facter.value(:python3_version) }```

if you need more python facts you can contribute them to our python module :slightly_smiling_face:

I didnt even think to look for python module..

imho, it's not a fact :slightly_smiling_face: one can have infinitive amount of pythons installed :slightly_smiling_face:

It would be great if you could open an issue over at the pdk repo 

And include your modules gemfile along with the dependencies in your metadata.json 

You should only see an issue like this if you are including gems that haven’t been packaged with PDK 

In general though I’d like to work out a way for people to be able to compile native gems if they wish 

Afaik the vendored mingw doesn’t carry the required files for compilation 

Good Morning <@U0332SZQLKZ> Thanks for the response...  If you could point me to where I can open an issue for the PDK repo...I will gladly do so.  As a side note, I have a lot of non control-repo modules which are mainly Heira data...Would love to be able to make that an option to help in configuring the pdk for a module... I feel like I have a ton of Gems in the Gemfile which are not necessary because of lack of functionality int he particular repository.  I also want to request or get help in creating a Job in CD4PE for these kind of repos which perform the Hiera syntax validate which are outside the control-repo.

One more thing... you made an interesting note in requesting the metadata.json...  Is there documentation on how that should relate to the Gemfile.,..  For this particular repo, after running PDK convert .. here are the last three lines of the metadata.json...  "pdk-version": "2.6.1",
  "template-url": "pdk-default#1.18.1",
  "template-ref": "tags/1.18.1-0-g3d2e75c"       I feel like this is a mix of the two version ...  1.18.1 was the old version and 2.6.1 is the one it is going to... Trying to understand if this is an issue or not.

I would recommend updating PDK to the latest version

PDKs repo is here <https://github.com/puppetlabs/pdk>

stick all the info you can in there and we can try to replicate for you

Your message also prompted me to do some reading around mingw64 and how we could allow people to have the support of native gems on windows

quite a big topic as you can probably imagine but very interesting

Hey Craig.... What are my options, such that even after upgrading to 2.7.1, pdk validate does not work because yet another Gem does not get successfully installed (rainbow this time).  Even if I run gem install rainbow --version '2.2.2' and that successfully completes... when I run pdk validate, it complains that the bundler cannot install rainbow.  Confusing... Yes.  Do I look into standing up a linux vm and use the pdk there?  or am I missing something?

You can also access PDKs bundled with

pdk bundle

So for a gem install you would do

pdk bundle exec gem install …

You can also use that to see what gems are available 

Did you get round to opening an issue? It would be really useful to see if your gemfile differs from the default one 

Hi Craig... took a fresh copy of the repo and reran commands to ensure I did not do something stupid :wink:  Opened <https://github.com/puppetlabs/pdk/issues/1244>  for this issue.

Is there any documentation on running pdk bundle exec gem commands...  I apologize for my lack of experience with ruby and gems... but I do learn a whole lot every day.

there are different tools to validate of the syntax is correct. puppet validate, metadata-json-lint, pylint, ruby

there are linters to ensure you follow the style guide and use best practices, for example rubocop and puppet-lint

there are unit test with rspec-puppet to test parts of the catalog

there are acceptance tests with beaker and serverspec to validate puppet provides systems how you expect it

there is puppet-catalog_diff to compare catalogs between different servers/module versions/puppet version to check the impact of code changes before it's deployed to prod

See the `puppet-catalog_diff` module at <https://forge.puppet.com/puppet/catalog_diff?src=slack&amp;channel=puppet-enterprise>

usually my customers implement it in this order (and not everything of it)

I am looking for something that does what CD4PE does as an alternative really, I was thinking jenkins.

jenkins works, gitlab is easier in my opinion. if you're on github you can use their actions

thankyou , i did wonder if that was a sensible option

I have an issue with Code Manager . running r10k manually as suggested <https://www.puppet.com/docs/pe/2019.8/code_mgr_troubleshoot.html#troubleshooting_code_manager-run-deployment-test> works fine

Support ticket? Or something easy I can check?

can you share the output? Like, does `puppet code` timeout when speaking to code manager, or does code manager timewhen when talking to the control repo git server or to the forg/other modules?

there are a fine timeout limits that can be configured