This is an old revision of the document!
Subversion (SVN) uses a centralized model to manage software projects. Collaborators check the source code out from the repository to their local development environments, make changes, then check code back in.
The repository can be on the local machine or a remote server, as SVN supports several protocols (svn://, http:// using Apache's mod_dav_svn module, file://, etc)..
$ sudo aptitude install subversion
Configure Apache:
Create the project skeleton (the name isn't important, it is only a folder in the file system):
$ sudo mkdir /tmp/myproject $ sudo mkdir /tmp/myproject/{branches,tags,trunk}
Create the SVN repository, :
$ sudo mkdir /mnt/repositories/myproject $ sudo svnadmin create /mnt/repositories/myproject
Import the initial project skeleton:
$ svn import /tmp/myproject/ file:///mnt/repositories/myproject/ -m "Initial import of project skeleton"
Clean up temp files and repository permissions. The repository needs www-data
permissions because Apache will be serving our repository for us:
$ sudo rm -rf /tmp/myproject $ sudo find /mnt/repositories/myproject/ -type f -exec chmod 660 {} \; $ sudo find /mnt/repositories/myproject/ -type d -exec chmod 2770 {} \; $ sudo chown -R root.www-data /mnt/repositories/myproject/
Tell Apache to use /svn/myproject
URLs with the dav_svn module, /etc/apache2/conf.d/svn.conf
:
<Location /svn/myproject> DAV svn SVNPath /mnt/repositories/myproject AuthzSVNAccessFile /mnt/repositories/authz_svn_access AuthName "Microsoft Active Directory Authentication" AuthType Basic PerlAuthenHandler Apache2::AuthenMSAD PerlSetVar MSADDomain ilri.cgiarad.org PerlSetVar MSADServer 172.26.0.218 # require any valid AD user to access the repo via Apache, # then use fine-grained controls in the AuthzSVNAccessFile # above. See: http://svnbook.red-bean.com/nightly/en/svn.serverconfig.pathbasedauthz.html#id646000 require valid-user # Or, use specific (but dumb) controls here in the Apache # config file instead of the AuthzSVNAccessFile above. #require user aorth akihara </Location>
Gracefully restart Apache to enable the changes:
$ sudo apache2ctl graceful
Check out the development branch ("trunk") to a local folder ("myproject"):
$ cd code/ $ svn co --username=aorth https://172.26.17.2/svn/beca/trunk myproject
Import your existing code base ("mycode") into the project:
$ cd myproject $ cp -R /home/aorth/mycode/* . $ svn add * $ svn commit -m "Initial import of codebase"
Sometimes you need to ignore certain files/folders (logs, temp files, etc) so that SVN doesn't try to commit those changes when it finds new files. For example: ignore everything in our project's tmp
folder:
$ svn propedit svn:ignore tmp/
And then place a single "*" in the text editor to ignore everything in the tmp folder. Otherwise, you can do something like "*.log" to ignore all log files.
$ cd myproject/ $ svn copy . http://172.26.17.2/svn/myproject/tags/0.2
Switch a working copy to a certain release:
$ svn sw http://172.26.17.2/svn/myproject/tags/0.2
Create a new Trac project:
$ sudo trac-admin /var/lib/trac/myproject initenv
Change the permissions so the Apache user can manage the Trac environment:
$ sudo chown -R www-data:www-data /var/lib/trac/myproject
Configure Apache to use Python for '/trac
' URLs. Put settings in a new configuration file, /etc/apache2/conf.d/trac.conf
:
<Location /trac> SetHandler mod_python PythonHandler trac.web.modpython_frontend PythonOption TracEnvParentDir /var/lib/trac PythonOption TracUriRoot /trac </Location>
The TracEnvParentDir
option tells it to allow access to all Trac instances in the specified directory.
Add Active Directory support for login URLs:
<Location /trac/myproject/login> AuthName "Microsoft Active Directory Authentication" AuthType Basic PerlAuthenHandler Apache2::AuthenMSAD PerlSetVar MSADDomain ilri.cgiarad.org PerlSetVar MSADServer 172.26.0.218 #require valid-user require user aorth akihara jmagochi </Location>
This will tie Active Directory authentication into the login system for Trac. We use the same scheme for SVN too, so users will be able to log in as the same user they are submitting code as.
Add a user as an admin in a particular Trac project:
$ sudo trac-admin /var/lib/trac/myproject permission add aorth TRAC_ADMIN
If your Trac gets out of sync with your SVN repository you will have to resync them:
$ sudo trac-admin /var/lib/trac/myproject resync Resyncing repository history... 1 revision cached. Done.
# aptitude install build-essential rake ruby1.8 ruby1.8-dev irb1.8 rdoc1.8 zlib1g-dev libopenssl-ruby1.8 libzlib-ruby libssl-dev rubygems1.8
Problems I faced:
You SHOULD be able to gem install
the modules, but I downloaded them all manually and installed manually. It turns out it was a DNS issue, and I had to add some entries to the machine's /etc/hosts file:
216.137.39.169 production.cf.rubygems.org d2chzxaqi4y7f8.cloudfront.net 72.21.202.165 production.s3.rubygems.org
Don't ask. I guess they are using the "cloud" and Amazon to host the files. I found these IPs from doing DNS queries outside ILRI.
$ wget http://rubygems.org/downloads/rack-1.0.1.gem $ wget http://rubygems.org/downloads/rails-2.3.5.gem $ wget http://rubygems.org/downloads/mysql-2.8.1.gem $ wget http://rubygems.org/downloads/sqlite3-ruby-1.3.0.gem $ wget http://rubygems.org/downloads/rake-0.8.7.gem $ wget http://rubygems.org/downloads/fastthread-1.0.7.gem
Rake, for building gems:
# gem install rake-0.8.3.gem
Make sure you can execute the rake
binary:
# cd /usr/local/bin/ # ln -s /var/lib/gems/1.8/bin/rake
Continue installing gems:
# gem install rack-1.0.1.gem # gem install fastthread-1.0.7.gem # gem install mysql-2.8.1.gem # gem install sqlite3-ruby-1.3.0.gem # gem install rails-2.3.5.gem
Download and unzip the latest stable tarball:
$ wget http://rubyforge.org/frs/download.php/70486/redmine-0.9.4.tar.gz $ tar zxf redmine-0.9.4.tar.gz $ cd redmine-0.9.4
We can either use MySQL or SQLite. SQLite has less overhead, and I'm not sure if it's any slower so we'll use that.
Edit config/database.yml:
production: adapter: sqlite3 dbfile: db/redmine.db
$ chmod 600 config/database.yml $ rake rake generate_session_store $ RAILS_ENV=production rake db:migrate $ RAILS_ENV=production rake redmine:load_default_data
Before we can use Redmine with Apache, we have to make sure the installation worked fine. Use the built-in webserver to test first:
$ script/server -e production
If you get an error about not finding ruby, make sure the ruby version in the script/server
script is correct.
Navigate to http://ip:3000 to see Redmine working. The default username and password are "admin"
Phusion Passenger ("mod_rails") is an Apache module which allows Apache to run ruby code (like mod_perl, mod_php, etc).
# gem install passenger
Edit the invocation line of the passenger installer program to point to ruby1.8
:
#!/usr/bin/env ruby1.8
Run the installer:
# /var/lib/gems/1.8/gems/passenger-2.2.14/bin/passenger-install-apache2-module
Apache will serve Redmine via FastCGI using the included dispatch.fcgi
script. This is much faster than using Webrick or regular CGI.
Rename it and make sure its executable:
$ mv public/dispatch.fcgi.example public/dispatch.fcgi $ chmod 755 public/dispatch.fcgi
Edit the invocation line in the FastCGI script to make sure it points to your ruby version (ruby1.8
), and that the dispatcher
line is correct:
#!/usr/bin/env ruby1.8 ... require "/var/lib/gems/1.8/gems/rails-2.3.5/lib/dispatcher.rb" ...
Change permissions so the web server can read/write important files/folders:
$ chown -R www-data:www-data /opt/redmine-0.9.4
Link the Redmine installation to the web document root:
# ln -s /opt/redmine-0.9.4/public/ /var/www/redmine