Fork me on GitHub

Metaparticle/Package for Python Tutorial

This is an in-depth tutorial for using Metaparticle/Package for Python

For a quick summary, please see the about Metaparticle.

Initial Setup

Check the tools

The docker command line tool needs to be installed and working. Try: docker ps to verify this. Go to the install page if you need to install Docker.

The mp-compiler command line tool needs to be installed and working. Try mp-compiler --help to verify this. Go to the releases page if you need to install the Metaparticle compiler.

Install Metaparticle/Package

Install the library

pip install metaparticle_pkg

Get the tutorial code

git clone https://github.com/metaparticle-io/package
cd package/tutorials/python

Initial Program

Inside of the tutorials/python directory, you will find a simple python project.

from six.moves import SimpleHTTPServer, socketserver
import socket

OK = 200

port = 8080

class MyHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
    def do_GET(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()
        self.wfile.write("Hello Metaparticle [{}] @ {}\n".format(self.path, socket.gethostname()).encode('UTF-8'))
        print("request for {}".format(self.path))
    def do_HEAD(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()

def main():
    Handler = MyHandler
    httpd = socketserver.TCPServer(("", port), Handler)
    httpd.serve_forever()

if __name__ == '__main__':
    main()

Step One: Containerize the Application

To build a container from our simple application we need to add a dependency to our code and then update the code to read as follows:

from six.moves import SimpleHTTPServer, socketserver
import socket
from metaparticle import containerize

OK = 200

port = 8080

class MyHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
    def do_GET(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()
        self.wfile.write("Hello Metaparticle [{}] @ {}\n".format(self.path, socket.gethostname()).encode('UTF-8'))
        print("request for {}".format(self.path))
    def do_HEAD(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()

@containerize(
    'docker.io/your-docker-user-goes-here', options={'name': 'my-image', 'publish': True})
def main():
    Handler = MyHandler
    httpd = socketserver.TCPServer(("", port), Handler)
    httpd.serve_forever()

if __name__ == '__main__':
    main()

You will notice that we added a @containerize annotation that describes how to package the application. You will need to replace your-docker-user-goes-here with an actual Docker repository path.

You can run this new program with:

python web.py

This code will start your web server again. But this time, it is running inside a container. You can see this by running:

docker ps

Step Two: Exposing the ports

If you try to access the web server on http://localhost:8080 you will see that you can not actually access the server. Despite it running, the service is not exposed. To do this, you need to add am annotation to supply the port(s) to expose.

The code snippet to add is:

...
@containerize('docker.io/your-docker-user-goes-here', options={'ports': ['8080']})
...

This tells the runtime the port(s) to expose. The complete code looks like:

import SimpleHTTPServer
import SocketServer
import socket
from metaparticle import containerize

OK = 200

port = 8080

class MyHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
    def do_GET(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()
        self.wfile.write("Hello Metaparticle [{}] @ {}\n".format(self.path, socket.gethostname()))
        print("request for {}".format(self.path))
    def do_HEAD(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()

@containerize(
    'docker.io/your-docker-user-goes-here',
    options={
        'ports': [8080],
        'name': 'my-image',
        'publish': True
    })
def main():
    Handler = MyHandler
    httpd = SocketServer.TCPServer(("", port), Handler)
    httpd.serve_forever()

if __name__ == '__main__':
    main()

Now if you run this with python web.py your webserver will be successfully exposed on port 8080.

Replicating and exposing on the web.

As a final step, consider the task of exposing a replicated service on the internet. To do this, we’re going to expand our usage of the @containerize tag. First we will add a replicas field, which will specify the number of replicas. Second we will set our execution environment to metaparticle which will launch the service into the currently configured Kubernetes environment.

Here’s what the snippet looks like:

...
@containerize(
    'docker.io/your-docker-user-goes-here',
    options={
        'replicas': 4,
        'executor': 'metaparticle',
        'ports': [8080],
        'name': 'my-image',
        'publish': True
    })
...

And the complete code looks like:

import SimpleHTTPServer
import SocketServer
import socket
from metaparticle import containerize

OK = 200

port = 8080

class MyHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
    def do_GET(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()
        self.wfile.write("Hello Metaparticle [{}] @ {}\n".format(self.path, socket.gethostname()))
        print("request for {}".format(self.path))
    def do_HEAD(self):
        self.send_response(OK)
        self.send_header("Content-type", "text/plain")
        self.end_headers()

@containerize(
    'docker.io/your-docker-user-goes-here',
    options={
        'ports': [8080],
        'replicas': 4,
        'runner': 'metaparticle',
        'name': 'my-image',
        'publish': True
    })
def main():
    Handler = MyHandler
    httpd = SocketServer.TCPServer(("", port), Handler)
    httpd.serve_forever()

if __name__ == '__main__':
    main()

After you run this, you can see that there are four replicas running behind a Kubernetes Service Load balancer:

$ kubectl get pods
...
$ kubectl get services
...