Install an existing application

On this page we explain all the procedures to build an application from only the code.

For example If you want to use an existing database you should ignore all the commands concerning the database.

This guide considers that:
  • We use a server manages by Camptocamp, meaning:
    • all dependencies described in the requirements are installed,
    • Postgres has a GIS template template_postgis and a user www-data,
    • Apache uses the user www-data.
  • Use Git as revision control.

For the others system there is some notes to give some help.

Set up the database

Any c2cgeoportal application requires a PostgreSQL/PostGIS database. The application works with its own tables, which store users, layers, etc. These tables are located in a specific schema of the database.


Multiple specific schemas are actually used in a parent/child architecture.

If the application has MapServer layers linked to PostGIS tables, these tables and the application-specific tables must be in the same database, preferably in separate schemas. This is required for layer access control (restricted layers), where joining user/role tables to PostGIS layer tables is necessary.

Create the database

To create the database you can use:

sudo -u postgres createdb <db_name> -T template_postgis

with <db_name> replaced by the actual database name.


if you do not have the template_postgis you can use:

with Postgres >= 9.1 and PostGIS >= 2.1:

sudo -u postgres createdb -E UTF8 -T template0 <db_name>
sudo -u postgres psql -c "CREATE EXTENSION postgis;" <db_name>

with older versions:

sudo -u postgres createdb -E UTF8 -T template0 <db_name>
sudo -u postgres psql -d <db_name> -f /usr/share/postgresql/9.1/contrib/postgis-2.1/postgis.sql
sudo -u postgres psql -d <db_name> -f /usr/share/postgresql/9.1/contrib/postgis-2.1/spatial_ref_sys.sql

Note that the path of the postgis scripts and the template name can differ on your host.

Create the schema

Each parent or child needs two application-specific schemas, then to create it use:

sudo -u postgres psql -c "CREATE SCHEMA <schema_name>;" <db_name>
sudo -u postgres psql -c "CREATE SCHEMA <schema_name>_static;" <db_name>

with <db_name> and <schema_name> replaced by the actual database name, and schema name (‘main’ by default), respectively.

Create a database user

We use a specific user for the application, www-data by default.


It the user does not already exist in your database, create it first:

sudo -u postgres createuser -P <db_user>

Give the rights to the user:

sudo -u postgres psql -c 'GRANT SELECT ON TABLE spatial_ref_sys TO "www-data"' <db_name>
sudo -u postgres psql -c 'GRANT ALL ON TABLE geometry_columns TO "www-data"' <db_name>
sudo -u postgres psql -c 'GRANT ALL ON SCHEMA <schema_name> TO "www-data"' <db_name>
sudo -u postgres psql -c 'GRANT ALL ON SCHEMA <schema_name>_static TO "www-data"' <db_name>


If you do not use the www-data user for Apache replace it by the right user.

Install the application

Get the application source tree

If Git is used for the application use the following command to get the application source tree:

git clone<project>.git
cd <project>

Non Apt/Dpkg based OS Configuration

For example Windows or RedHat.

Disable the package checking:

In the <package>.mk add:


Windows Specific Configuration

Some Python modules cannot currently be installed through the Python Package Index (pypi) and they have to be downloaded manually and stored. This is because these packages use DLLs and binaries which would have to be compiled using a C compiler.

Furthermore, some changes in the apache wsgi and mapserver configurations are required to make c2cgeoportal work on Windows.

Also, between all the different command interfaces available on Windows (cmd, Cygwin, git mingw), only Windows default cmd interface handle paths correctly in all stage of the application setup.

Command interface and environment variable

Only use Windows default commande interface:

Start > Run... > cmd

Cygwin and git mingw are not compatible. Powershell is untested.

Complementarily you need to add all the ressource paths to your system PATH environment variable, for cygwin, git and node binaries.


You must install the following packages:

  • make
  • git
  • gettext-devel

Python Wheels

You should create a “wheels” folder at the root folder of your project.

Then, go to, search and download the following packages:

  • Psycopg2
  • Shapely
  • Pillow
  • Pyproj

If your project is configured for Windows, then make will expect this folder to exist and to contain these wheels.

To be sure to use the right version of these packages, open the CONST_requirements_windows.txt file and modify the versions of these packages according to the files you have downloaded.


WSGIDaemonProcess and WSGIProcessGroup are not supported on windows.

(WSGIDaemonProcess ConfigurationDirective “Note that the WSGIDaemonProcess directive and corresponding features are not available on Windows or when running Apache 1.3.”)

The following lines must be commented/removed:

WSGIDaemonProcess c2cgeoportal:${instanceid} display-name=%{GROUP} user=${modwsgi_user}
WSGIProcessGroup c2cgeoportal:${instanceid}


The path to Mapserver executable must be modified:

ScriptAlias /${instanceid}/mapserv C:/path/to/ms4w/Apache/cgi-bin/mapserv.exe


You must specify the path to the mapserver’s epsg file by uncommenting and adapting this line under MAP (use regular slash /)

PROJ_LIB" "C:/PATH/TO/ms4w/proj/nad"


The following configuration override must be added to your <project>.mk:

# Sets that is we use Windows
# Path to cygwin
CYGWIN_PATH ?= c:/path/to/cygwin
APACHE_VHOST ?= your_apache_vhost_folder
APACHE_CONF_DIR ?= path/to/your/$(APACHE_VHOST)/conf
PRINT_OUTPUT ?= path/to/your/Tomcat7/webapps
TOMCAT_SERVICE_COMMAND ?= path/to/your/Tomcat7/bin/Tomcat7.exe
APACHE_GRACEFUL ?= path/to/your/Apache/bin/httpd.exe -k restart -n <servicename>
# Where <servicename> is the name of the Apache service, look at your
# Windows services panel (Start > Search > Services)

RedHat Specific Configuration

Specific settings are required when the c2cgeoportal application is to be run on RedHat Enterprise Linux (RHEL) 6.


First of all, note that, with RHEL, you cannot install the c2cgeoportal application in your homedir. If you do so, you will get the following error in the Apache logs:

(13)Permission denied: access to /~elemoine/ denied

So always install the application in an Apache-accessible directory. On Camptocamp puppetized servers you will typically install the application in /var/www/vhosts/<vhost>/private/dev/<username>/, where <vhost> is the name of the Apache virtual host, and <username> is your Unix login name.


Configure some <package>.mk RedHat specifics:

# Service base command for Tomcat on RedHat
TOMCAT_SERVICE_COMMAND ?= sudo service tomcat-tomcat1
# Command prefix to acces to the webapp folder
TOMCAT_OUTPUT_CMD_PREFIX ?= sudo -u tomcat
# Unix user who execute the 'mod_wsgi' processes.
# apache on RedHat, www-data elsewhere
MODWSGI_USER ?= apache
# Graceful command for RedHat
APACHE_GRACEFUL ?= sudo /usr/sbin/apachectl graceful
# Test packages are designed for Debian. They must be disabled by adding the following
# line to ``<package>.mk``::
DISABLE_BUILD_RULES += test-packages test-packages-tilecloud-chain test-packages-mobile test-packages-ngeo


On RHEL 6 the mapserv binary is located in /usr/libexec/. The mapserver.conf.mako Apache config file assumes that mapserv is located in /usr/lib/cgi-bin/, and should therefore be changed:

ScriptAlias /${instanceid}/mapserv /usr/libexec/mapserv


Ensure that the regular expression used in apache/application.wsgi.mako to modify the sys.path matches the system directories containing python packages. If you are installing from scratch, this should already be the case; otherwise look at scaffolds/create/apache/application.wsgi.mako for an example.

Install the application

If it does not already exist, create a <user>.mk file (where <user> is for example your username), that will contain your application special configuration:

INSTANCE_ID = <instanceid>

include <package>.mk


For technical simplification the file should be named <instanceid>.mk in other words <instanceid> should have the same value as <user>.


The <instanceid> should be unique on the server, the username is a good choice or something like <user>-<sub-project> in case of parent/children project.

Add it to Git:

git add <user>.mk
git commit -m "Add user build file"

Create the application tables, and directly set the version (details later):

make -f <user>.mk upgrade-db

Then you can build and install the application with the command:

make -f <user>.mk build

This previous command will do many things like:

  • download and install the project dependencies,
  • adapt the application configuration to your environment,
  • build the javascript and css resources into compressed files,
  • compile the translation files.

Your application should be available at: http://<hostname>/<instanceid>.

Where the <hostname> is directly linked to the virtual host, and the <instanceid> is the value you provided before.

Migrating to a new server

If you are migrating to a new server, keep in mind that your variable VISIBLE_WEB_HOST must contain the exact host name that browsers should use to access your site. Consider the following migration scenario: your current site runs on server with the visible host name You wish to setup a new server, install the application and test it, and then switch your DNS so that now points to To accomplish this, you must proceed as follows:

  • install the application on and test the application at
  • later, when going live, you must:
    • change VISIBLE_WEB_HOST to
    • re-build, re-deploy - but do not test yet!
    • change your DNS so that points to
    • Now test your new live site.