$ rhc app-create mypythonapp python-2.7
By default, OpenShift comes with all 3 versions of Python and the pattern of building and deploying applications is the same for all three. The Python cartridges are based on a WSGI server model of Python web creation. This tutorial will start with showing you how to create an application using the default, Apache with Mod_WSGI, but it will also show you to use CherryPy as an example of using a different WSGI server.
This tutorial walks through the creation of an application titled mypythonapp under the domain mydomain. We will use Python 2.7 but you can substitute in Python 2.6 or 3.3 and use them in the same way. Before going through the rest of the tutorial, make sure to have created an OpenShift account and installed the RHC command line tools.
Once that is completed, open a terminal on a local machine and change into a directory where the source code will be located. Then at the command prompt enter the following command to create a Python application:
$ rhc app-create mypythonapp python-2.7
The output will look something like this:
Application Options ------------------- Domain: mydomain Cartridges: python-2.7 Gear Size: default Scaling: no Creating application 'mypythonapp' ... done Waiting for your DNS name to be available ... done Cloning into 'mypythonapp'... Warning: Permanently added 'mypythonapp-mydomain.rhcloud.com,184.108.40.206' (RSA) to the list of known hosts. Your application 'mypythonapp' is now available. URL: http://mypythonapp-mydomain.rhcloud.com/ (1) SSH to: firstname.lastname@example.org Git remote: ssh://email@example.com/~/git/mypythonapp.git/ Cloned to: /home/me/git_demo/mypythonapp (2) Run 'rhc show-app mypythonapp' for more details about your app.
|1||Public URL of application|
|2||Location of local Git repository|
The application will now be running and accessible at the URL (1) listed in the output. A local git repository for your application was also created in the directory listed under Cloned to (2).
A developer on OpenShift can make code changes on a local machine, check those changes in locally, and then
push those changes to OpenShift. One of the primary advantages of Git is that it does not require a continuous online presence in order to run. A developer can easily check in (in Git terminology,
commit) and revert changes locally before deciding to upload those changes to OpenShift.
To make the first change, edit
wsgi.py in the base of the local git repository. Go to the file in the terminal and edit with a command line editor like VIM or EMacs. Alternatively, edit the file using any text editor or IDE. The beginning of this file starts a basic WSGI application with some URL routing. Strip out everything between the two
<body> tags and change the content to look like this:
<body> <h1>Hello world!</h1> </body>
Now that you have made this incredibly complicated code change, the code must be deployed to the server. Accomplish this task using Git.
To add the change to the Git repository’s index and stage it for the next commit, use the command
$ git add --all .
commit the staged changes to the Git repository, with a log message to explain the changes:
$ git commit -m "Your commit message"
push these changes to the remote OpenShift application.
$ git push
The output of the push command will first show the git push and then the build process and deployment for the OpenShift application. When the build is finished, go back to the application’s URL (mypythonapp-mydomain.rhcloud.com) and see a page which just says "Hello World" in nice big font.
This is the basic process to follow when developing an application:
git commit, and finally
If you are curious about Git or would like to learn more, the Pro Git book is a great resource.
When you look in the git repository you will notice there is both a setup.py and a requirements.txt. Just as in a standard Python application, both of these files can be used to manage package dependencies. We recommend that you use setup.py to hold the metadata for your application, such as name and author, and requirements.txt for the specific versions of packages your application needs. This follows the documentation for setup scripts and an excellent blog piece by one of the main pip maintainers. Here is the documentation for requirements.txt.
OpenShift runs pip install -r requirements.txt on build and adds the packages to the applications virtualenv. If you are unfamiliar with virtualenv you should read some of the documentation. For Python 3.3 applications the proper way to load the virtualenv should look like this:
virtenv = os.environ['OPENSHIFT_PYTHON_DIR'] + '/virtenv/' virtualenv = os.path.join(virtenv, 'bin/activate_this.py') try: # See: http://stackoverflow.com/questions/23418735/using-python-3-3-in-openshifts-book-example?noredirect=1#comment35908657_23418735 #execfile(virtualenv, dict(__file__=virtualenv)) # for Python v2.7 #exec(compile(open(virtualenv, 'rb').read(), virtualenv, 'exec'), dict(__file__=virtualenv)) # for Python v3.3 # Multi-Line for Python v3.3: exec_namespace = dict(__file__=virtualenv) with open(virtualenv, 'rb') as exec_file: file_contents = exec_file.read() compiled_code = compile(file_contents, virtualenv, 'exec') exec(compiled_code, exec_namespace) except IOError: pass
As you can see there are instructions here to make this work in Python 2.7.
As noted above, by default OpenShift uses Mod_WSGI as the default WSGI server but it is quite easy to use a different WSGI server. Let’s go ahead and start CherryPy instead of Mod_WSGI.
In the requirements.txt be sure to add the dependecy on CherryPy:
You need to create a file named app.py in the base of your local repository. Don’t forget to add it to the git index by entering:
$ git add app.py
Remove the virtualenv statements that you find in wsgi.py and add them to the top of this file. If you are using Python 3m please make sure it matches the statements under managing dependencies.
Add code to start your WSGI server. Please be aware you need to use the OpenShift environment variables that give you the IP and port.
Load up the app class, which in this example will be in wsgi.py.
Here is the code in app.py:
#!/usr/bin/python import os import sys import wsgi from cherrypy import wsgiserver #hack to make sure we can load wsgi.py as a module in this class sys.path.insert(0, os.path.dirname(__file__)) virtenv = os.environ['OPENSHIFT_PYTHON_DIR'] + '/virtenv/' virtualenv = os.path.join(virtenv, 'bin/activate_this.py') try: #execfile(virtualenv, dict(__file__=virtualenv)) # for Python v2.7 #exec(compile(open(virtualenv, 'rb').read(), virtualenv, 'exec'), dict(__file__=virtualenv)) # for Python v3.3 # Multi-Line for Python v3.3: exec_namespace = dict(__file__=virtualenv) with open(virtualenv, 'rb') as exec_file: file_contents = exec_file.read() compiled_code = compile(file_contents, virtualenv, 'exec') exec(compiled_code, exec_namespace) except IOError: pass # Get the environment information we need to start the server ip = os.environ['OPENSHIFT_PYTHON_IP'] port = int(os.environ['OPENSHIFT_PYTHON_PORT']) host_name = os.environ['OPENSHIFT_GEAR_DNS'] server = wsgiserver.CherryPyWSGIServer((ip, port), wsgi.application, server_name=host_name) server.start()
For wsgi.py you just need to remove the code that loads the virtual environment and it will work like before.
While there are standard environment variables that can help to manage your Python application, there are also some which are specifc to just Python. You can use them to manage the entry-point for WSGI or where your requirements files is located. They are discussed more fully in their own document. They can be helpful when you have your own format for a git repository or you are using a framework, such as Django, that expects files in a certain location.
With a normal
git push as outlined above, Openshift starts and stops the Python server on each build. For Python applications, your code can be built and deployed without restarting the server. See Hot Deployment for more information on how OpenShift uses a marker in the git repository to turn on this build style. Please be aware that there may be times, such as loading a new library, that you want to restart the server. Either rename or remove the file from your git repository to get the server to restart.
You can also force the rebuild of the virtual environment by adding a force_clean_build into your git repo under .openshift/markers. Don’t forget to add this to your git index.
Find out how to add a database to your application by going to the Adding a Database guide.
|You should only use environment variables to specify the connection parameters for your database. Using hard coded names, ports, or credentials limits the resusability of your app and can potentially break your app during OpenShift maintenance.|
The best next step is to create an application using OpenShift.
Browse our quickstarts and community cartridges to see other exciting technology you can use in your applications.
Finally, if at any point you have questions or issues, please visit the Help Center for a full list of options.