If you’re running a recent version of Ubuntu or Fedora, you can use the ‘quickinstall’ script to set up a quick Molly instance to play with.
Note that the quickinstall script will override any existing Postgres configuration you may have, so you are recommended to run this on a server where Postgres only runs Molly (or, you can do it the hard way below).
First off, you will need a checkout of Molly. You can either grab a tarball from the Downloads part of Github, and unzip that:
wget --no-check-certificate https://github.com/mollyproject/mollyproject/tarball/master
tar zxvf https://github.com/mollyproject/mollyproject/tarball/master
Or clone the Git repository directly:
git clone git://github.com/mollyproject/mollyproject.git
(note that you may have to install Git for this to work)
The links above will get the latest “unstable” version of Molly, you may wish to select a specific version from the download page, or checkout a specific branch from Git (e.g., git checkout molly1.0).
Once you have a copy of the Molly source on your system, you can just run the quickinstaller:
sudo mollyproject/quickinstall /path/to/whereever
or if you’re on a Red Hat-like system:
su -c "mollyproject/quickinstall /path/to/whereever"
Where /path/to/whereever is where you would like to install Molly (e.g., /srv/molly)
The quick installer will take you through the process of installing and configuring Molly, leaving you with a fully functional install in your destination. In the destination there will also be a folder called ‘deploy’ - this will contain your site and any customisations to Molly you have. You will probably want to copy this somewhere else and keep it under version control. By default, it only consists of a settings file, a manage file, a sample Apache config and default directories for your customised media (site_media) and templates (templates). Please note that the media/ folder is used to contain the compiled media, and will be automatically populated by the installer - do not edit the files in here!
To start up the development server for Django for your newly installed site, then simply run the following:
cd /path/to/molly
source bin/activate
cd deploy
python manage.py runserver
This will start a default server on localhost, port 8000 (it will give you the IP as a startup message), until you press Ctrl+C. You can then use this server to start the batch jobs to populate your initial database. Optionally, you can start the server to listen to the outside world by altering the last step to:
python manage.py runserver 0.0.0.0:8000
The Django webserver is not sufficient to run a production copy of Molly, therefore it is recommended that you use Apache and mod_wsgi to host the site. The sample site created by the installer consists of a WSGI script and a sample Apache config in the apache/ directory of your site.
The mod_wsgi documentation goes into considerable detail about how to deploy a Django application.
Once you have started working on your Molly-based site, you will undoubtedly want to keep the code in some sort of version control system, separate from the main Molly code. For example, to set up a deployment on a new machine, or updating an existing install.
The -s option can be passed to the quickinstall script to tell it to use an already existing site, rather than create a new one. For example, if you have a Subversion checkout of your site in /home/me/site, and a copy of the Molly code at /home/me/mollyproject, and Molly installed at /srv/molly, you can update your Molly install to use your site by running:
sudo /home/me/mollyproject/quickinstall -s /home/me/site /srv/molly
To use the quick install script to install Molly in “development mode” (which means that installation takes place quicker by missing out some optional steps, such as database configuration, and that the code is run directly from your source checkout, rather than being installed), simply pass the -d flag to quickinstall, e.g.:
./quickinstall -d /home/chris/molly/
Note that you do not need to run as root to install Molly in development mode.
Because a development install does not check pre-requisites (or configure the database if you need it to do so), you will be required to install Molly’s system requisites by hand (check ‘The Hard Way’ below), or to first do a non-development installation to ensure the system wide pre-requisites are satisfied.
To update Molly using the quick-installer, you simply need to run the quickinstall script from the new version pointing it to the same location. This will recognise that the installation already exists and will update as appropriate.
When you are asked for Molly’s database name, if you created the database using the quick installer, this will default to ‘molly’.
If you can’t, or don’t want to, use the Molly quick installer, you can install Molly by hand as a standard Python package.
First off, you’ll need to install the pre-requisites for Molly and its dependencies. On Fedora, this list is:
On Ubuntu (and Debian), this is:
Other distributions have similar package names
You need to decide how you would like Molly to be installed - globally on the system, or to a directory using a Python virtualenv. If you decide to install it into a Python virtualenv, then you can create one like so:
virtualenv --distribute --no-site-packages PATH-TO-DIRECTORY
source PATH-TO-DIRECTORY/bin/activate
(note, if you’re installing Molly globally, some systems call “pip” “pip-python”, so you may need to use that instead)
Molly uses forks of 2 packages in PyPI so these must be manually installed first:
pip install -U git+http://github.com/oucs/PyZ3950.git#egg=PyZ3950
pip install -U git+git://github.com/mikelim/django-compress.git#egg=django-compress
And sometimes PIL fails to install when being installed as a dependency, so:
pip install -U PIL
Now you are ready to install Molly. This can be done by running:
python setup.py install
in the Molly directory. This will install Molly in your environment, as well as pulling in any further dependencies, and you’re almost ready to go.
The next step is to prepare the Postgres database for Molly. You will first need to create a spatial template (the Django documentation covers how to do this), and then use this to create a database for Molly:
sudo -u postgres createdb -T template_postgis molly
You’ll then need to create a user for Molly to connect as:
sudo -u postgres createuser molly -P
You will be asked to enter a password, and you can choose ‘n’ for the remaining questions. You will also need to configure Postgres for MD5 based authentication. This can be done by adding the following line:
local molly molly md5
to the pg_hba.conf (in /var/lib/pgsql/data on Fedora/Red Hat, /etc/postgresql/8.4/main on Ubuntu/Debian) - above the lines ending “ident”.
Finally, you’ll need to give the molly user access to the molly database:
sudo -u postgres psql -c "GRANT ALL ON DATABASE molly TO molly;"
Once this is done, you’re now ready to configure Molly. If you don’t already have a Molly site, you will need to create one to do this, this can be done by running:
django-admin createproject site
Which will create a site in a folder called “site”. You will then need to configure this to use Molly (the configuring guide may help you, or alternatively the config_generator.py script in the scripts/ folder will generate a typical settings.py).
Once you have a site, you need to update the database. If you’re installing into a blank database, run the following commands in the site folder:
python manage.py syncdb --all
python manage.py migrate --fake
You should also be asked to create a superuser. Never run those commands on an instantiated database though, otherwise it may become out of sync. If your database is not blank, then run:
python manage.py syncdb
python manage.py migrate
At this point, you should also make sure that the cache directory exists, and is writeable to the user Molly is running as.
You can now build Molly’s media:
python manage.py update_wurfl
python manage.py collectstatic
python manage.py synccompress
python manage.py generate_cache_manifest
python manage.py generate_markers
And, if you would like the batch jobs to run when scheduled:
python manage.py create_crontab | crontab
And tada! Molly is installed and ready to go (so configure Apache, or run the development server: python manage.py runserver)