The mod_wsgi package provides an Apache module that implements a WSGI compliant interface for hosting Python based web applications on top of the Apache web server.
Installation of mod_wsgi from source code can be performed in one of two ways.
The first way of installing mod_wsgi is the traditional way that has been
used by many software packages. This is where it is installed as a module
directly into your Apache installation using the commands configure
,
make
and make install
, a method sometimes referred to by the
acronym CMMI. This method works with most UNIX type systems. It cannot
be used on Windows.
The second way of installing mod_wsgi is to install it as a Python package
into your Python installation using the Python pip install
command.
This can be used on all platforms, including Windows.
This second way of installing mod_wsgi will compile not only the Apache module for mod_wsgi, but will also install a Python module and admin script, which on UNIX type systems can be used to start up a standalone instance of Apache directly from the command line with an auto generated configuration.
This later mechanism for installing mod_wsgi using Python pip
is a much
simpler way of getting starting with hosting your Python web application.
In particular, this installation method makes it very easy to use
Apache/mod_wsgi in a development environment without the need to perform
any Apache configuration yourself.
The Apache module for mod_wsgi created when using the pip install
method can still be used with the main Apache installation, via manual
configuration if necessary. As detailed later in these instructions, the
admin script installed when you use pip install
can be used to generate
the configuration to manually add to the Apache configuration to load
mod_wsgi.
Note that although MacOS X is a UNIX type system, the pip install
method is the only supported way for installing mod_wsgi.
Since MacOS X Sierra, Apple has completely broken the ability to install
third party Apache modules using the apxs
tool normally used for this
task. History suggests that Apple will never fix the problem as they have
broken things in the past in other ways and workarounds were required as
they never fixed those problems either. This time there is no easy
workaround as they no longer supply certain tools which are required to
perform the installation.
The pip install
method along with the manual configuration of Apache
is also the method you need to use on Windows.
With either installation method for mod_wsgi, you must have Apache installed. This must be a complete Apache installation. It is not enough to have only the runtime packages for Apache installed. You must have the corresponding development package for Apache installed, which contains the Apache header files, as these are required to be able to compile and install third party Apache modules.
Similarly with Python, you must have a complete Python installation which includes the corresponding development package, which contains the header files for the Python library.
If you are running Debian or Ubuntu Linux with Apache 2.2 system packages, and were using the Apache prefork MPM you would need both:
- apache2-mpm-prefork
- apache2-prefork-dev
If instead you were using the Apache worker MPM, you would need both:
- apache2-mpm-worker
- apache2-threaded-dev
If you are running Debian or Ubuntu Linux with Apache 2.4 system packages, regardless of which Apache MPM is being used, you would need both:
- apache2
- apache2-dev
If you are running RHEL, CentOS or Fedora, you would need both:
- httpd
- httpd-devel
If you are using the Software Collections Library (SCL) packages with RHEL, CentOS or Fedora, you would need:
- httpd24
- httpd24-httpd-devel
If you are running MacOS X, Apache is supplied with the operating system.
If running a recent MacOS X version, you will though need to have the Xcode
command line tools installed as well as the Xcode application. The command
line tools can be installed by running xcode-select --install
. The
Xcode application can be installed from the MacOS X App Store. If you are
using older MacOS X versions, you may be able to get away with having just
the command line tools.
If you are running Windows, it is recommended you use the Apache
distribution from Apache Lounge (www.apachelounge.com). Other Apache
distributions for Windows aren't always complete and are missing the files
needed to compile additional Apache modules. By default, it is expected
that Apache is installed in the directory C:/Apache24
on Windows.
For installation directly into your Apache installation using the CMMI method, see the full documentation at:
Alternatively, use the following instructions to install mod_wsgi into your Python installation and then either copy the mod_wsgi module into your Apache installation, or configure Apache to use the mod_wsgi module from the Python installation.
When using this approach, you will still need to manually configure Apache to have mod_wsgi loaded into Apache, and for it to know about your WSGI application.
To install mod_wsgi directly into your Python installation, from within the source directory of the mod_wsgi package you can run:
python setup.py install
This will compile mod_wsgi and install the resulting package into your Python installation.
If wishing to install an official release direct from the Python Package Index (PyPi), you can instead run:
pip install mod_wsgi
If you wish to use a version of Apache which is installed into a non
standard location, you can set and export the APXS
environment variable
to the location of the Apache apxs
script for your Apache installation
before performing the installation.
If you are on Windows and your Apache distribution is not installed into
the directory C:/Apache24
, first set the environment variable
MOD_WSGI_APACHE_ROOTDIR
to the directory containing the Apache
distribution. Ensure you use forward slashes in the directory path. The
directory path should not include path components with spaces in the name.
Note that nothing will be copied into your Apache installation at this point. As a result, you do not need to run this as the root user unless installing it into a site wide Python installation rather than a Python virtual environment. It is recommended you always use Python virtual environments and never install any Python package directly into the system Python installation.
On a UNIX type system, to verify that the installation was successful, run
the mod_wsgi-express
script with the start-server
command:
mod_wsgi-express start-server
This will start up Apache/mod_wsgi on port 8000. You can then verify that the installation worked by pointing your browser at:
http://localhost:8000/
When started in this way, the Apache web server will stay in the foreground. To stop the Apache server, use CTRL-C.
For a simple WSGI application contained in a WSGI script file called
wsgi.py
, in the current directory, you can now run:
mod_wsgi-express start-server wsgi.py
This instance of the Apache web server will be completely independent of, and will not interfere with any existing instance of Apache you may have running on port 80.
If you already have another web server running on port 8000, you can
override the port to be used using the --port
option:
mod_wsgi-express start-server wsgi.py --port 8080
For a complete list of options you can run:
mod_wsgi-express start-server --help
For further information related to using mod_wsgi-express
see the main
mod_wsgi documentation.
Many Linux distributions have a tendency to screw around with the standard
Apache Software Foundation layout for installation of Apache. This can
include renaming the Apache httpd
executable to something else, and in
addition to potentially renaming it, replacing the original binary with a
shell script which performs additional actions which can only be performed
as the root
user.
In the case of the httpd
executable simply being renamed, the
executable will obviously not be found and mod_wsgi-express
will fail
to start at all.
In this case you should work out what the httpd
executable was renamed
to and use the --httpd-executable
option to specify its real location.
For example, if httpd
was renamed to apache2
, you might need to use:
mod_wsgi-express start-server wsgi.py --httpd-executable=/usr/sbin/apache2
In the case of the httpd
executable being replaced with a shell script
which performs additional actions before then executing the original
httpd
executable, and the shell script is failing in some way, you will
need to use the location of the original httpd
executable the shell
script is in turn executing.
The primary intention of mod_wsgi-express
is to make it easier for
users to run up Apache on non privileged ports, especially during the
development of a Python web application. If you want to be able to run
Apache using mod_wsgi-express
on a privileged port such as the standard
port 80 used by HTTP servers, then you will need to run
mod_wsgi-express
as root. In doing this, you will need to perform
additional steps.
The first thing you must do is supply the --user
and --group
options to say what user and group your Python web application should run
as. Most Linux distributions will predefine a special user for Apache to
run as, so you can use that. Alternatively you can use any other special
user account you have created for running the Python web application:
mod_wsgi-express start-server wsgi.py --port=80 \ --user www-data --group www-data
This approach to running mod_wsgi-express
will be fine so long as you
are using a process supervisor which expects the process being run to remain
in the foreground and not daemonize.
If however you are directly integrating into the system init scripts where
separate start and stop commands are expected, with the executing process
expected to be daemonized, then a different process is required to setup
mod_wsgi-express
.
In this case, instead of simply using the start-server
command to
mod_wsgi-express
you should use setup-server
:
mod_wsgi-express setup-server wsgi.py --port=80 \ --user www-data --group www-data \ --server-root=/etc/mod_wsgi-express-80
In running this command, it will not actually startup Apache. All it will do is create the set of configuration files and the startup script to be run.
So that these are not created in the default location of a directory under
/tmp
, you should use the --server-root
option to specify where they
should be placed.
Having created the configuration and startup script, to start the Apache instance you can now run:
/etc/mod_wsgi-express-80/apachectl start
To subsequently stop the Apache instance you can run:
/etc/mod_wsgi-express-80/apachectl stop
You can also restart the Apache instance as necessary using:
/etc/mod_wsgi-express-80/apachectl restart
Using this approach, the original options you supplied to setup-server
will be cached with the same configuration used each time. If you need to
update the set of options, run setup-server
again with the new set of
options.
Note that even taking all these steps, it is possible that running up
Apache as root
using mod_wsgi-express
may fail on systems where
SELinux extensions are enabled. This is because the SELinux profile may not
match what is being expected for the way that Apache is being started, or
alternatively, the locations that Apache has been specified as being
allowed to access, don't match where the directory specified using the
--server-root
directory was placed. You may therefore need to configure
SELinux or move the directory used with --server-root
to an allowed
location.
In all cases, any error messages will be logged to a file under the server
root directory. If you are using mod_wsgi-express
with a process
supervisor, or in a container, where log messages are expected to be sent
to the terminal, you can use the --log-to-terminal
option.
To use mod_wsgi-express
with Django, after having installed the
mod_wsgi package into your Python installation, edit your Django settings
module and add mod_wsgi.server
to the list of installed apps.
INSTALLED_APPS = ( 'django.contrib.admin', 'django.contrib.auth', 'django.contrib.contenttypes', 'django.contrib.sessions', 'django.contrib.messages', 'django.contrib.staticfiles', 'mod_wsgi.server', )
To prepare for running mod_wsgi-express
, ensure that you first collect
up any Django static file assets into the directory specified for them in
the Django settings file:
python manage.py collectstatic
You can now run the Apache server with mod_wsgi hosting your Django application by running:
python manage.py runmodwsgi
If working in a development environment and you would like to have any code
changes automatically reloaded, then you can use the --reload-on-changes
option.
python manage.py runmodwsgi --reload-on-changes
If wanting to have Apache started as root in order to listen on port 80,
instead of using mod_wsgi-express setup-server
as described above,
use the --setup-only
option to the runmodwsgi
management command.
python manage.py runmodwsgi --setup-only --port=80 \ --user www-data --group www-data \ --server-root=/etc/mod_wsgi-express-80
This will set up all the required files and you can use apachectl
to
start and stop the Apache instance as explained previously.
If you want to use mod_wsgi in combination with your system Apache installation, the CMMI method for installing mod_wsgi would normally be used.
If you are on MacOS X Sierra that is no longer possible. Even prior to MacOS X Sierra, the System Integrity Protection (SIP) system of MacOS X, prevented installing the mod_wsgi module into the Apache modules directory.
If you are using Windows, the CMMI method was never supported as Windows doesn't supply the required tools to make it work.
The CMMI installation method also involves a bit more work as you need to
separately download the mod_wsgi source code, run the configure
tool
and then run make
and make install
.
The alternative to using the CMMI installation method is to use the Apache
mod_wsgi module created by running pip install
. This can be directly
referenced from the Apache configuration, or copied into the Apache modules
directory.
To use the Apache mod_wsgi module from where pip install
placed it,
run the command mod_wsgi-express module-config
. This will output
something like:
LoadModule wsgi_module /usr/local/lib/python2.7/site-packages/mod_wsgi/server/mod_wsgi-py27.so WSGIPythonHome /usr/local/lib
These are the directives needed to configure Apache to load the mod_wsgi module and tell mod_wsgi where the Python installation directory or virtual environment was located.
This would be placed in the Apache httpd.conf
file, or if using a Linux
distribution which separates out module configuration into a
mods-available
directory, in the wsgi.load
file within the
mods-available
directory. In the latter case where a mods-available
directory is used, the module would then be enabled by running
a2enmod wsgi
as root
. If necessary Apache can then be restarted to
verify the module is loading correctly. You can then configure Apache as
necessary for your specific WSGI application.
Note that because in this scenario the mod_wsgi module for Apache could be
located in a Python virtual environment, if you destroy the Python virtual
environment the module will also be deleted. In that case you would need to
ensure you recreate the Python virtual environment and reinstall the
mod_wsgi package using pip
, or, take out the mod_wsgi configuration
from Apache before restarting Apache, else it will fail to startup.
Instead of referencing the mod_wsgi module from the Python installation,
you can instead copy the mod_wsgi module into the Apache installation. To
do that, run the mod_wsgi-express install-module
command, running it as
root
if necessary. This will output something like:
LoadModule wsgi_module modules/mod_wsgi-py27.so WSGIPythonHome /usr/local/lib
This is similar to above except that the mod_wsgi module was copied to the
Apache modules directory first and the LoadModule
directive references
it from that location. You should take these lines and configure Apache in
the same way as described above.
Do note that copying the module like this will not work on recent versions of MacOS X due to the SIP feature of MacOS X.