|Version 337 (modified by 9 years ago) ( diff ),|
This page documents the 1.4 (latest stable) release. Documentation for other releases can be found here.
Trac Installation Guide for 0.12
Table of Contents
Since version 0.12, Trac can also be localized, and there's probably a translation available for your language. If you want to be able to use the Trac interface in other languages, then make sure you first have installed the optional package Babel. Lacking Babel, you will only get the default english version, as usual. If you install Babel later on, you will need to re-install Trac.
If you're interested in contributing new translations for other languages or enhance the existing translations, then please have a look at TracL10N.
What follows are generic instructions for installing and setting up Trac and its requirements. While you may find instructions for installing Trac on specific systems at TracInstallPlatforms on the main Trac site, please be sure to first read through these general instructions to get a good understanding of the tasks involved.
- Installing Trac
- Creating a Project Environment
- Deploying Trac
- Configuring Authentication
- Finishing the install
To install Trac, the following software packages must be installed:
- Python, version ≥ 2.4 and < 3.0 (note that we dropped the support for Python 2.3 in this release)
- setuptools, version ≥ 0.6
- Genshi, version ≥ 0.6
You also need a database system and the corresponding python bindings. The database can be either SQLite, PostgreSQL or MySQL.
For the SQLite database
If you're using Python 2.5 or 2.6, you already have everything you need.
If you're using Python 2.4 and need pysqlite, you can download from google code the Windows installers or the tar.gz archive for building from source:
$ tar xvfz <version>.tar.gz $ cd <version> $ python setup.py build_static install
This will extract the SQLite code and build the bindings.
To install SQLite, your system may require the development headers. Without these you will get various GCC related errors when attempting to build:
$ apt-get install libsqlite3-dev
SQLite 2.x is no longer supported, and neither is PySqlite 1.1.x.
A known bug PySqlite versions 2.5.2-4 prohibits upgrade of trac databases from 0.11.x to 0.12. Please use versions 2.5.5 and newer or 2.5.1 and older. See #9434 for more detail.
See additional information in PySqlite.
For the PostgreSQL database
You need to install the database and its Python bindings:
See DatabaseBackend for details.
For the MySQL database
Trac can now work quite well with MySQL, provided you follow the guidelines.
It is very important to read carefully the MySqlDb page before creating the database.
Version Control System
- Subversion, 1.5.x or 1.6.x and the corresponding Python bindings. Older versions starting from 1.4.0, etc. should still work. For troubleshooting information, check the TracSubversion page. Versions prior to 1.4.0 won't probably work since trac uses svn core functionality (e.g. svn_path_canonicalize) that is not implemented in the python swig wrapper in svn ⇐ 1.3.x (although it exists in the svn lib itself).
There are pre-compiled SWIG bindings available for various platforms. (Good luck finding precompiled SWIG bindings for any Windows package at that listing. TracSubversion points you to Algazam, which works for me under Python 2.6.)
Note that Trac doesn't use PySVN, neither does it work yet with the newer
ctype-style bindings. [Is there a ticket for implementing ctype bindings?]
Please note: if using Subversion, Trac must be installed on the same machine. Remote repositories are currently not supported.
A web server is optional because Trac is shipped with a server included, see the Running the Standalone Server section below.
Alternatively you configure Trac to run in any of the following environments.
- Apache with
- mod_wsgi, see TracModWSGI and http://code.google.com/p/modwsgi/wiki/IntegrationWithTrac
- mod_python 3.3.1, deprecated: see TracModPython)
- a FastCGI-capable web server (see TracFastCgi)
- an AJP-capable web server (see TracOnWindowsIisAjp)
- a CGI-capable web server (see TracCgi), but usage of Trac as a cgi script is highly discouraged, better use one of the previous options.
Other Python Packages
- Babel, version ≥ 0.9.5,
needed for localization support
Note: If you want to be able to use the Trac interface in other languages, then make sure you first have installed the optional package Babel. Lacking Babel, you will only get the default english version, as usual. If you install Babel later on, you will need to re-install Trac.
- docutils, version ≥ 0.3.9 for WikiRestructuredText.
- Pygments for syntax highlighting. SilverCity and/or Enscript may still be used but are deprecated and you really should be using Pygments.
- pytz to get a complete list of time zones, otherwise Trac will fall back on a shorter list from an internal time zone implementation.
Attention: The various available versions of these dependencies are not necessarily interchangable, so please pay attention to the version numbers above. If you are having trouble getting Trac to work please double-check all the dependencies before asking for help on the MailingList or IrcChannel.
Please refer to the documentation of these packages to find out how they are best installed. In addition, most of the platform-specific instructions also describe the installation of the dependencies. Keep in mind however that the information there probably concern older versions of Trac than the one you're installing (there are even some pages that are still talking about Trac 0.8!).
One way to install Trac is using setuptools. With setuptools you can install Trac from the subversion repository;
A few examples:
- first install of the latest stable version Trac 0.12.1, with i18n support:
easy_install Babel==0.9.5 Genshi==0.6 easy_install TracIt's very important to run the two
easy_installcommands separately, otherwise the message catalogs won't be generated.
- upgrade to the latest stable version of Trac:
easy_install -U Trac
- upgrade to the latest trunk development version (0.13dev):
easy_install -U Trac==dev
For upgrades, reading the TracUpgrade page is mandatory, of course.
If you want more control, you can download the source in archive form, or do a checkout from one of the official source code repositories.
Be sure to have the prerequisites already installed. You can also obtain the Genshi and Babel source packages from http://www.edgewall.org and follow for them a similar installation procedure, or you can just easy_install those, see above.
Once you've unpacked the Trac archive or performed the checkout, move in the top-level folder and do:
$ python ./setup.py install
You'll need root permissions or equivalent for this step.
This will byte-compile the python source code and install it as an .egg file or folder in the
of your Python installation. The .egg will also contain all other resources needed by standard Trac, such as htdocs and templates.
If you install from source and want to make Trac available in other languages, make sure Babel is installed. Only then, perform the
install (or simply redo the
install once again afterwards if you realize Babel was not yet installed):
$ python ./setup.py install
Alternatively, you can do a
bdist_egg and copy the .egg from dist/ to the place of your choice, or you can create a Windows installer (
To install Trac to a custom location, or find out about other advanced installation options, run:
Also see Installing Python Modules for detailed information.
Specifically, you might be interested in:
or, if installing Trac to a Mac OS X system:
easy_install --prefix=/usr/local --install-dir=/Library/Python/2.5/site-packages
Note: If installing on Mac OS X 10.6 running
easy_install http://svn.edgewall.org/repos/trac/trunk will install into
/Library/Python/2.6/site-packages by default
The above will place your
trac-admin commands into
/usr/local/bin and will install the Trac libraries and dependencies into
/Library/Python/2.5/site-packages, which is Apple's preferred location for third-party Python application installations.
'pip' is an easy_install replacement that is very useful to quickly install python packages. To get a trac installation up and running in less than 5 minutes:
Assuming you want to have your entire pip installation in /opt/user/trac:
pip -E /opt/user/trac install trac psycopg2
pip -E /opt/user/trac install trac mysql-python
Make sure your OS specific headers are available for pip to automatically build PostgreSQL (libpq-dev) or MySQL (libmysqlclient-dev) bindings.
pip will automatically resolve all dependencies (like Genshi, pygments, etc.) and download the latest packages on pypi.python.org and create a self contained installation in /opt/user/trac .
All commands (tracd, trac-admin) are available in /opt/user/trac/bin. This can also be leveraged for mod_python (using PythonHandler directive) and mod_wsgi (using WSGIDaemonProcess directive)
Additionally, you can install several trac plugins (listed here) through pip.
Creating a Project Environment
A Trac environment is the backend storage where Trac stores information like wiki pages, tickets, reports, settings, etc. An environment is basically a directory that contains a human-readable configuration file, and various other files and directories.
A new environment is created using trac-admin:
$ trac-admin /path/to/myproject initenv
trac-admin will prompt you for the information it needs to create the environment, such as the name of the project and the database connection string. If you're not sure what to specify for one of these options, just press
<Enter> to use the default value.
Using the default database connection string in particular will always work as long as you have SQLite installed. For the other database backends you should plan ahead and already have a database ready to use at this point.
Since 0.12, Trac doesn't ask for a source code repository anymore when creating an environment. Repositories can be added afterward, or the version control support can be disabled completely if you don't need it.
Also note that the values you specify here can be changed later by directly editing the conf/trac.ini configuration file.
Finally, make sure the user account under which the web front-end runs will have write permissions to the environment directory and all the files inside. This will be the case if you run
trac-admin ... initenv as this user. If not, you should set the correct user afterwards. For example on Linux, with the web server running as user
apache and group
# chown -R apache.apache /path/to/myproject
Warning: Please only use ASCII-characters for account name and project path, unicode characters are not supported there.
Running the Standalone Server
After having created a Trac environment, you can easily try the web interface by running the standalone server tracd:
$ tracd --port 8000 /path/to/myproject
Then, fire up a browser and visit
http://localhost:8000/. You should get a simple listing of all environments that
tracd knows about. Follow the link to the environment you just created, and you should see Trac in action. If you only plan on managing a single project with Trac you can have the standalone server skip the environment list by starting it like this:
$ tracd -s --port 8000 /path/to/myproject
Running Trac on a Web Server
Trac provides various options for connecting to a "real" web server:
- mod_python (no longer recommended, as mod_python is not actively maintained anymore)
- CGI (should not be used, as the performance is far from optimal)
Generating the Trac cgi-bin directory
In order for Trac to function properly with FastCGI you need to have a
trac.fcgi file and for mod_wsgi a
trac.wsgi file. These are Python scripts which load the appropriate Python code. They can be generated using the
deploy option of trac-admin.
There is, however, a bit of a chicken-and-egg problem. The trac-admin command requires an existing environment to function, but complains if the deploy directory already exists. This is a problem, because environments are often stored in a subdirectory of the deploy. The solution is to do something like this:
mkdir -p /usr/share/trac/projects/my-project trac-admin /usr/share/trac/projects/my-project initenv trac-admin /usr/share/trac/projects/my-project deploy /tmp/deploy mv /tmp/deploy/* /usr/share/trac
Mapping Static Resources
Out of the box, Trac will pass static resources such as style sheets or images through itself. For anything but a tracd only based deployment, this is far from optimal as the web server could be set up to directly serve those static resources (for CGI setup, this is highly undesirable and will cause abysmal performance).
Web servers such as Apache allow you to create “Aliases” to resources, giving them a virtual URL that doesn't necessarily reflect the layout of the servers file system. We also can map requests for static resources directly to the directory on the file system, avoiding processing these requests by Trac itself.
There are two primary URL paths for static resources -
/chrome/site. Plugins can add their own resources usually accessible by
/chrome/plugin path, so its important to override only known paths and not try to make universal
/chrome alias for everything.
Note that in order to get those static resources on the filesystem, you need first to extract the relevant Trac resources using the
deploy command of TracAdmin:
deploy <directory> Extract static resources from Trac and all plugins
htdocs directory within the target
Example: Apache and
Add the following snippet to Apache configuration before the
WSGIScriptAlias (which map all the other requests to the Trac application), changing paths to match your deployment:
Alias /trac/chrome/common /path/to/trac/htdocs/common Alias /trac/chrome/site /path/to/trac/htdocs/site <Directory "/path/to/www/trac/htdocs"> Order allow,deny Allow from all </Directory>
If using mod_python, you might want to add this too (otherwise, the alias will be ignored):
<Location "/trac/chrome/common/"> SetHandler None </Location>
Note that we mapped
/trac part of the URL to the
trac.*cgi script, and the path
/chrome/common is the path you have to append to that location to intercept requests to the static resources.
Similarly, if you have static resources in a project's
htdocs directory (which is referenced by /chrome/site URL in themes), you can configure Apache to serve those resources (again, put this before the
WSGIScriptAlias for the .*cgi scripts, and adjust names and locations to match your installation):
Alias /trac/chrome/site /path/to/projectenv/htdocs <Directory "/path/to/projectenv/htdocs"> Order allow,deny Allow from all </Directory>
Alternatively to aliasing
/trac/chrome/site, you can tell Trac to generate special links for those static resources, using the [trac] htdocs_location configuration setting:
[trac] htdocs_location = http://static.example.org/trac-htdocs
Note that this makes it easy to have a dedicated domain serve those static resources (preferentially cookie-less).
Of course, you still need to make the Trac
htdocs/common directory available through the web server at the specified URL, for example by copying (or linking) the directory into the document root of the web server:
$ ln -s /path/to/www/trac/htdocs /var/www/yourhost.example.org/trac-htdocs
Setting up the Plugin Cache
Some Python plugins need to be extracted to a cache directory. By default the cache resides in the home directory of the current user. When running Trac on a Web Server as a dedicated user (which is highly recommended) who has no home directory, this might prevent the plugins from starting. To override the cache location you can set the PYTHON_EGG_CACHE environment variable. Refer to your server documentation for detailed instructions on how to set environment variables.
Trac uses HTTP authentication. You'll need to configure your webserver to request authentication when the
.../login URL is hit (the virtual path of the "login" button). Trac will automatically pick the REMOTE_USER variable up after you provide your credentials. Therefore, all user management goes through your web server configuration. Please consult the documentation of your web server for more info.
The process of adding, removing, and configuring user accounts for authentication depends on the specific way you run Trac.
We'll describe here the most common scenario.
Example: Basic Authentication with Apache
The simplest way to enable authentication with Apache is to create a password file. Use the
htpasswd program to create the password file:
$ htpasswd -c /somewhere/trac.htpasswd admin New password: <type password> Re-type new password: <type password again> Adding password for user admin
After the first user, you dont need the "-c" option anymore:
$ htpasswd /somewhere/trac.htpasswd john New password: <type password> Re-type new password: <type password again> Adding password for user john
See the man page for
htpasswdfor full documentation.
After you've created the users, you can set their permissions using TracPermissions.
Now, you'll need to enable authentication against the password file in the Apache configuration:
<Location "/trac/login"> AuthType Basic AuthName "Trac" AuthUserFile /somewhere/trac.htpasswd Require valid-user </Location>
If you're hosting multiple projects you can use the same password file for all of them:
<LocationMatch "/trac/[^/]+/login"> AuthType Basic AuthName "Trac" AuthUserFile /somewhere/trac.htpasswd Require valid-user </LocationMatch>
Example: Digest Authentication with Apache
For better security, it is recommended that you either enable SSL or at least use the “digest” authentication scheme instead of “Basic”. Please read the Apache HTTPD documentation to find out more. For example, on a Debian 4.0r1 (etch) system the relevant section in apache configuration can look like this:
<Location "/trac/login"> LoadModule auth_digest_module /usr/lib/apache2/modules/mod_auth_digest.so AuthType Digest AuthName "trac" AuthDigestDomain /trac AuthUserFile /somewhere/trac.htpasswd Require valid-user </Location>
and you'll have to create your .htpasswd file with htdigest instead of htpasswd as follows:
# htdigest /somewhere/trac.htpasswd trac admin
where the "trac" parameter above is the same as AuthName above ("Realm" in apache-docs).
More authentication scenarios
To learn more how to setup authentication for the frontend you're using, please refer to one of the following pages:
- TracStandalone if you use the standalone server,
- TracModWSGI if you use the Apache mod_wsgi web front end.
- TracModPython if you use the Apache mod_python web front end.
Finishing the install
Automatic reference to the SVN changesets in Trac tickets
You can configure SVN to automatically add a reference to the changeset into the ticket comments, whenever changes are committed to the repository. The description of the commit needs to contain one of the following formulas:
Refs #123- to reference this changeset in
Fixes #123- to reference this changeset and close
#123ticket with the default status fixed
This functionality requires a post-commit hook to be installed as described in TracRepositoryAdmin, and enabling the optional commit updater components by adding the following line to the
[components] section of your trac.ini, or enabling the components in the "Plugins" admin panel.
tracopt.ticket.commit_updater.* = enabled
For more information, see the documentation of the
CommitTicketUpdater component in the "Plugins" admin panel.
Once you have your Trac site up and running, you should be able to create tickets, view the timeline, browse your version control repository if configured, etc.
Keep in mind that anonymous (not logged in) users can by default access only a few of the features, in particular they will have a read-only access to the resources. You will need to configure authentication and grant additional permissions to authenticated users to see the full set of features.