Docker Compose is a tool for defining and
running multi-container Docker applications. With compose you define a
set of containers, along with their relationships, in a YAML file. And
then use the docker-compose CLI to launch and manage them.
For this example we've provided a compose file using the version 2 schema. This means you will need at least Docker Engine 1.10 and Compose 1.6.
The example launches a stack running Puppet Server, PuppetDB, a
PostgresDB container for PuppetDB and the open source dashboards
Puppetboard and Puppet Explorer. See the docker-compose.yml file for
details.
Run docker-compose up to launch the containers. Note that you'll see
lots of output as the various software launches in parallel.
$ docker-compose up
Creating compose_puppetboard_1
Creating postgres
Creating puppet
Creating compose_puppetexplorer_1
...
One the output from compose stops, in another terminal confirm you have the containers up and running.
$ docker ps ~
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
7af731797a95 puppet/puppetexplorer "/usr/bin/caddy" 2 minutes ago Up 2 minutes 0.0.0.0:32828->80/tcp compose_puppetexplorer_1
8f8aecd2cc69 puppet/puppetserver "/opt/puppetlabs/bin/" 2 minutes ago Up 2 minutes 0.0.0.0:32827->8140/tcp puppet
1fefb0ea22ff puppet/puppetdb-postgres "/docker-entrypoint.s" 2 minutes ago Up 2 minutes 5432/tcp postgres
910002aa3e6a puppet/puppetboard "/usr/bin/gunicorn -b" 2 minutes ago Up 2 minutes 0.0.0.0:32824->8000/tcp compose_puppetboard_1
78bfd7967680 puppet/puppetdb "dumb-init /docker-en" 2 minutes ago Up 2 minutes 0.0.0.0:32826->8080/tcp, 0.0.0.0:32825->8081/tcp compose_puppetdb_1
Note the port number for the puppetexplorer container, in my case 32828. If
you're running docker locally you should be able to access the dashboard
on 0.0.0.0:32824 (your port will likely vary). If you're running
docker-machine then you can run docker-machine ip to determine the
correct IP address. So far the dashboard should be empty, we haven't run
any agents yet. Let's do that now.
You could connect any agents to the Puppet infrastructure launched above but for demonstration purposes we're just going to launch a few ephemeral puppet agents in Docker containers. These aren't useful in most sense, the container will run, receive instructions from Puppet and then disappear. But they are useful for testing and demonstration purposes.
Note that compose automatically creates a default network which we need to connect our containers to in order for them to find the Puppet Server.
Let's run up a few alpine agents. Run the following command a few times.
docker run --net compose_default puppet/puppet-agent-alpine
And then run some Ubuntu based agents. Again run the command a couple of times to populate the database.
docker run --net compose_default puppet/puppet-agent-ubuntu
Now access the dashboard described above again. You should have some data to explore.
PuppetDB also exposes a dashboard, showing various operational metrics,
as well as an API for
accessing all the collected resource data. You can find the port for the
dashboard using docker ps described above. The docker port command can
also be useful.
$ docker port compose_puppetdb_1
8080/tcp -> 0.0.0.0:32826
8081/tcp -> 0.0.0.0:32825
With that port in hand, and the ip address of the machine running docker, you can query the PuppetDB API.
$ curl -s -X GET http://192.168.99.100:32826/pdb/query/v4 --data-urlencode 'query=nodes {}' | jq
{
"deactivated": null,
"latest_report_hash": "f8332ac22e0abf6a51571fae6b57b2a881f207fe",
"facts_environment": "production",
"cached_catalog_status": "not_used",
"report_environment": "production",
"catalog_environment": "production",
"facts_timestamp": "2016-05-27T12:47:04.495Z",
"latest_report_noop": false,
"expired": null,
"report_timestamp": "2016-05-27T12:47:04.144Z",
"certname": "a9efc038b3ca",
"catalog_timestamp": "2016-05-27T12:47:05.038Z",
"latest_report_status": "changed"
},
{
"deactivated": null,
"latest_report_hash": "d273124e1e74708272228ac4465f6f1923100db7",
"facts_environment": "production",
"cached_catalog_status": "not_used",
"report_environment": "production",
"catalog_environment": "production",
"facts_timestamp": "2016-05-27T12:47:37.543Z",
"latest_report_noop": false,
"expired": null,
"report_timestamp": "2016-05-27T12:47:36.959Z",
"certname": "5a4cbf61e790",
"catalog_timestamp": "2016-05-27T12:47:38.050Z",
"latest_report_status": "changed"
}
]
Here I'm issuing a PQL query for all nodes. I'm parsing it through jq for nicer formatting.
PuppetDB stores a great deal of information, and PQL and the API provides a powerful way of accessing it. The Puppet-in-Docker setup makes for a great experimental test bed for building atop that capability.
In case you see errors like the this on the puppet container:
ERROR [puppetserver] Puppet Failed to execute '/pdb/cmd/v1?checksum=eeb40197db6a4ac3d8bce09778388cf7a812a621&version=5&certname=puppetdb.local&command=replace_facts' on at least 1 of the following 'server_urls': https://puppetdb:8081
ERROR [c.p.h.c.i.PersistentSyncHttpClient] Error executing http request
java.net.ConnectException: Connection refused
Try to uncomment the following lines in the definition of the puppet service in the docker-compose.yaml file:
environment:
- PUPPETDB_SERVER_URLS=https://puppetdb.local:8081
links:
- puppetdb:puppetdb.local