skeleton

Setup

$ pyvenv ./env
$ . env/bin/activate
$ pip install -r requirements.txt

Set up CoreOS with etcd

In order to test this POC, you will need to have CoreOS running so you can update config variables in etcd and have the application use them. I used Vagrant to run CoreOS across 3 nodes on a VM with VirtualBox.

Clone this project and follow the setup instuctions coreos-vagrant Once you have CoreOS up and running, you can set up the config structure with values to work with this project.

For my example, the application is skeleton and the environment is mthomas.

# create directories
$ curl -L http://172.17.8.101:2379/v2/keys/skeleton -XPUT -d dir=true
$ curl -L http://172.17.8.101:2379/v2/keys/skeleton/mthomas -XPUT -d dir=true

# create port and debug keys within skeleton/mthomas directory
$ curl -L http://172.17.8.101:2379/v2/keys/skeleton/mthomas/port -XPUT -d value=80
$ curl -L http://172.17.8.101:2379/v2/keys/skeleton/mthomas/debug -XPUT -d value=True

# See values for available environment
$ curl -L http://172.17.8.101:2379/v2/keys/skeleton/mthomas

Config Structure

The commands above set up the directory structure in the following format:

$ curl -L http://172.17.8.101:2379/v2/keys/skeleton/mthomas | python -m json.tool
{
    "action": "get",
    "node": {
        "createdIndex": 66250,
        "dir": true,
        "key": "/skeleton/mthomas",
        "modifiedIndex": 66250,
        "nodes": [
            {
                "createdIndex": 85833,
                "key": "/skeleton/mthomas/debug",
                "modifiedIndex": 85833,
                "value": "True"
            },
            {
                "createdIndex": 66289,
                "key": "/skeleton/mthomas/port",
                "modifiedIndex": 66289,
                "value": "80"
            }
        ]
    }
}

Run

Start up the app and pass in environment (defaults to mthomas).

Environment will need to correspond to an etcd subdirectory for the skeleton app.

The port will come from the etcd config!

$ python main.py --env=mthomas

Ping the app

Once the app is running, curl to it to make sure it's up.

(Replace 80 with a valid port from the etcd config)

$ curl -i http://localhost:80/ping

Publish to PyPi

The package version is set in skeleton/__init__.py.

In order to publish a package to PyPi, you will need to set up a pypirc file with credentials in your home directory. Here's a sample of what my pypirc looks like when publishing to a pypi server called internal.

[distutils]
index-servers =
  internal

[internal]
repository: http://someserver:8080
username: xxxxxxx
password: xxxxxxx

First, you need to publish the tarball for distribution.

$ python setup.py sdist

Next, you will need to register your PyPi server, in my case internal.

$ python setup.py register -r internal

Now you can publish the skeleton package (sdist) to PyPi.

$ python setup.py sdist upload -r internal

Install Package from PyPi

To install the skeleton package from the internal PyPi server, add the following line to the top of your requrements file. Replace http://someserver:8080 with the actual hostname of the PyPi server (specified in your ~/.pypirc).

-i http://someserver:8080/simple/