Puppet Orchestrator --application Example

10.26.2016

As always, this technical blog post began when I was having a problem and couldn't find the answer on the internet. I'm writing tests for Puppet's application orchestrator which involve setting up an application remotely then running some variations of puppet job run. In particular, I need to specify an environment (-e or --environment) and an application instance (-a or --application).

Here's my directory structure:

[root@hostname environments]# tree
.
├── production
│   ├── environment.conf
│   ├── hieradata
│   ├── manifests
│   │   └── site.pp
│   └── modules
└── tmpenv
    ├── manifests
    │   └── site.pp
    └── modules
        ├── app_one
        │   ├── manifests
        │   │   ├── db_component.pp
        │   │   ├── init.pp
        │   │   └── web_component.pp
        │   └── README.md
        ├── README.md
        └── shared_types_and_providers
            └── lib
                └── puppet
                    ├── provider
                    │   ├── testhttp
                    │   │   └── test_file.rb
                    │   └── testsql
                    │       └── test_file.rb
                    └── type
                        ├── testhttp.rb
                        └── testsql.rb

And the output of running puppet app show -e tmpenv is:

App_one[Health Check Test]
    App_one::Db_component[Health Check Test] => y21898u9zb3yin5.delivery.puppetlabs.net
        + produces Testsql[Health Check Test]
    App_one::Web_component[Health Check Test] => hzmjunpoua0mxeo.delivery.puppetlabs.net
        consumes Testsql[Health Check Test]

Ok, so, as I see it there are 2 strong candidates for what the "<APPLICATION>" referred to in the puppet orchestrator documentation could be: app_one or App_one[Health Check Test]. Here's the output for each of those (using puppet job plan so that I'm not creating new job id's and logs all over the place):

[root@hostname environments]# puppet job plan -e tmpenv -a app_one
Failed to submit plan: puppetlabs.orchestrator/unknown-target: The target was not found in environment tmpenv: app_one

[root@hostname environments]# puppet job plan -e tmpenv -a App_one[Health Check Test]
terminate called after throwing an instance of 'HorseWhisperer::action_validation_error' what():  Validation Error: You supplied too many arguments to the run command. Only 1 is allowed.
  Aborted

In desperation, I decide to try a combination of the two App_one.

[root@hostname environments]# puppet job plan -e tmpenv -a
App_one
+-------------------+---------+
| Environment       | tmpenv  |
| Target            | App_one |
| Concurrency Limit | None    |
| Nodes             | 2       |
+-------------------+---------+

Application instances: 1
  - App_one[Health Check Test]

Node run order (nodes in level 0 run before their dependent nodes in
level 1, etc.):
0 -----------------------------------------------------------------------
y21898u9zb3yin5.delivery.puppetlabs.net
    App_one[Health Check Test] - App_one::Db_component[Health Check
    Test]

1 -----------------------------------------------------------------------
hzmjunpoua0mxeo.delivery.puppetlabs.net
    App_one[Health Check Test] - App_one::Web_component[Health
    Check Test]

Use `puppet job run --application 'App_one' --environment
tmpenv` to create and run a job like this.
Node catalogs may have changed since this plan was
generated.

Success! I later discovered that puppet job plan -e tmpenv -a "App_one[Health Check Test]" also worked. It's still unclear to me why the name needs be capitalized, but I trust that there are good and well thought out Reasons which just aren't documented.

So, if you don't know, now you know :)