Difference between revisions of "Gitlab"

From ArchWiki
Jump to: navigation, search
(Create a virtual host for Gitlab)
(Enable host and start unicorn)
Line 94: Line 94:
Enable your Gitlab virtual host and reload [[Apache]]:
Enable your Gitlab virtual host and reload [[Apache]]:
{{hc|/etc/httpd/conf/httpd.conf|Include conf/vhosts/gitlab}}
{{hc|/etc/httpd/conf/httpd.conf| Include /etc/httpd/conf/extra/gitlab.conf}}
Finally start unicorn:
Finally start unicorn:

Revision as of 15:58, 4 December 2013

Template:Article summary start Template:Article summary text Template:Article summary heading Template:Article summary wiki Template:Article summary wiki Template:Article summary end Gitlab is a free git repository management application based on Ruby on Rails. It is distributed under the MIT License and its source code can be found on Github. It is a very active project with a monthly release cycle and ideal for businesses that want to keep their code private. Consider it as a self hosted Github but open source. You can try a demo here.

Note: Throughout the article, sudo is heavily used, assuming that the user that is running the commands is root or someone with equal privileges. There is no need to edit the sudoers file whatsoever. It is only used to change to the appropriate user. For more info read man sudo.


Simply install the gitlabAUR package from the AUR.

Note: In order to receive mail notifications, make sure to install a mail server. By default, Archlinux does not ship with one. The recommended mail server is postfix, but you can use others such as SSMTP, msmtp, sendmail, etc.
Note: If you want to use rvm be sure to check out Gitlab#Running GitLab with rvm before starting with the installation


Database backend

Currently GitLab supports MySQL and PostgreSQL. MariaDB has not been officially tested but it works just fine.


Install mariadb and libmariadbclient from the official repositories and start the daemon. Create the database and do not forget to replace your_password_here with a real one.

mysql -u root -p
mysql> CREATE DATABASE IF NOT EXISTS `gitlabhq_production`;
mysql> CREATE USER 'gitlab'@'localhost' IDENTIFIED BY 'your_password_here';
mysql> GRANT SELECT, LOCK TABLES, INSERT, UPDATE, DELETE, CREATE, DROP, INDEX, ALTER ON `gitlabhq_production`.* TO 'gitlab'@'localhost';
mysql> \q

Try connecting to the new database with the new user:

mysql -u gitlab -p -D gitlabhq_production


Install postgresql and libpqxx from the official repositories. Follow PostgreSQL#Installing_PostgreSQL to set it up and start the daemon. Login to PostgreSQL and remember to change your_password_here to a real one:

psql -d template1
template1=# CREATE USER gitlab WITH PASSWORD 'your_password_here';
template1=# CREATE DATABASE gitlabhq_production OWNER gitlab;
template1=# \q

Try connecting to the new database with the new user:

psql -d gitlabhq_production

Web server configuration

Nginx and unicorn

Install nginx from the official repositories.

Run these commands to setup nginx:

ln -s /etc/nginx/sites-available/gitlab /etc/nginx/sites-enabled/gitlab

Edit /etc/nginx/sites-enabled/gitlab and change YOUR_SERVER_IP and YOUR_SERVER_FQDN to the IP address and fully-qualified domain name of the host serving Gitlab. As you can see nginx needs to access /home/gitlab/gitlab/tmp/sockets/gitlab.socket socket file. You have to be able to run sudo -u http ls /home/gitlab/gitlab/tmp/sockets/gitlab.socket successfully. Otherwise setup access to the directory:

# chgrp http /home/gitlab
# chmod u=rwx,g=rx,o= /home/gitlab

Restart gitlab.service, resque.service and nginx.

Unicorn is an HTTP server for Rack applications designed to only serve fast clients on low-latency, high-bandwidth connections and take advantage of features in Unix/Unix-like kernels. First we rename the example file and then we start unicorn:

# cd /home/gitlab/gitlab
# sudo -u gitlab cp config/unicorn.rb.orig config/unicorn.rb
# sudo -u gitlab bundle exec unicorn_rails -c config/unicorn.rb -E production -D

Apache and unicorn

Install apache from the official repositories.

Configure Unicorn
Note: If the default path is not /home/git for your installation, change the below path accordingly

As the official installation guide instructs, copy the unicorn configuration file:

# sudo -u git -H cp /home/git/gitlab/config/unicorn.rb.example /home/git/gitlab/config/unicorn.rb

Now edit config/unicorn.rb and add a listening port by uncommenting the following line:

listen ""
Tip: You can set a custom port if you want. Just remember to also include it in Apache's virtual host. See below.
Create a virtual host for Gitlab

Create a configuration file for Gitlab’s virtual host and insert the lines below adjusted accordingly. For the ssl section see LAMP#SSL. If you do not need it, remove it. Notice that the SSL virtual host needs a specific IP instead of generic. Also if you set a custom port for Unicorn, do not forget to set it at the BalanceMember line.

Enable host and start unicorn

Enable your Gitlab virtual host and reload Apache:

 Include /etc/httpd/conf/extra/gitlab.conf

Finally start unicorn:

# cd /home/gitlab/gitlab
# sudo -u gitlab bundle exec unicorn_rails -c config/unicorn.rb -E production -D

Basic configuration

Open /etc/gitlab/gitlab.yml with your favorite editor and edit where needed. The options are pretty straightforward. Make sure to change localhost to the fully-qualified domain name of your host serving GitLab where necessary.

To configure GitLab database settings, make sure to update username/password in /etc/gitlab/database.yml. If you planning to use PostgreSQL backend, you should copy its template file before configuring it:

cp /usr/share/doc/gitlab/database.yml.postgresql /etc/gitlab/database.yml

Initialize Database

Note: Make sure the redis daemon is enabled and started, otherwise the following command will fail. To check the status and see if it's running execute systemctl status redis, if it's dead start it as per usual via systemctl start redis

Initialize database and activate advanced features:

$ bundle exec rake gitlab:setup RAILS_ENV=production

You have to run these commands as the gitlab user from the gitlab directory:

$ cd /usr/share/webapps/gitlab
$ sudo -u gitlab bundle exec rake gitlab:check RAILS_ENV=production
Note: If you recieve a error No such file or directory - /home/git/repositories/root then most likely you've changed the default configuration for GitLab and you'll need to modify all static paths in config/gitlab.yml and run the above command again to initialize the database!

Check status

With the following commands we check if the steps we followed so far are configured properly.

$ bundle exec rake gitlab:env:info RAILS_ENV=production
$ bundle exec rake gitlab:check RAILS_ENV=production
Example output of gitlab:env:info
System information
System:		Arch Linux
Current User:	git
Using RVM:	yes
RVM Version:	1.20.3
Ruby Version:	2.0.0p0
Gem Version:	2.0.0
Bundler Version:1.3.5
Rake Version:	10.0.4

GitLab information
Version:	5.2.0.pre
Revision:	4353bab
Directory:	/home/git/gitlab
DB Adapter:	mysql2
URL:		http://gitlab.arch
HTTP Clone URL:	http://gitlab.arch/some-project.git
SSH Clone URL:	git@gitlab.arch:some-project.git
Using LDAP:	no
Using Omniauth:	no

GitLab Shell
Version:	1.4.0
Repositories:	/home/git/repositories/
Hooks:		/home/git/gitlab-shell/hooks/
Git:		/usr/bin/git
Note: gitlab:check will complain about missing initscripts. Don't worry, we will use ArchLinux' systemd to manage server start (which GitLab does not recognize).

Start and test GitLab

$ systemctl daemon-reload

After starting the database backend, simply run:

$ systemctl start redis
$ systemctl start gitlab

To automatically launch GitLab at startup, run:

$ systemctl enable redis
$ systemctl enable gitlab

Useful Tips

Hook into /var

 sudo mkdir -m700 /var/log/gitlab /var/tmp/gitlab
 sudo chown gitlab:gitlab /var/log/gitlab /var/tmp/gitlab
 sudo -u gitlab -i
 cd ~/gitlab
 d=log; mv $d/* /var/$d/gitlab; rm -f $d/.gitkeep; rm -r $d && ln -s /var/$d/gitlab $d
 d=tmp; mv $d/* /var/$d/gitlab; rm -f $d/.gitkeep; rm -r $d && ln -s /var/$d/gitlab $d

Hidden options

Go to Gitlab's home directory

# cd /home/gitlab/gitlab

and run

# rake -T | grep gitlab

These are the options so far:

rake gitlab:app:backup_create      # GITLAB | Create a backup of the gitlab system
rake gitlab:app:backup_restore     # GITLAB | Restore a previously created backup
rake gitlab:app:enable_automerge   # GITLAB | Enable auto merge
rake gitlab:app:setup              # GITLAB | Setup production application
rake gitlab:app:status             # GITLAB | Check gitlab installation status
rake gitlab:gitolite:update_hooks  # GITLAB | Rewrite hooks for repos
rake gitlab:gitolite:update_keys   # GITLAB | Rebuild each key at gitolite config
rake gitlab:gitolite:update_repos  # GITLAB | Rebuild each project at gitolite config
rake gitlab:test                   # GITLAB | Run both cucumber & rspec

Backup and restore

Create a backup of the gitlab system:

# sudo -u gitlab -H rake RAILS_ENV=production gitlab:backup:create

Restore the previously created backup file /home/gitlab/gitlab/tmp/backups/20130125_11h35_1359131740_gitlab_backup.tar:

# sudo -u gitlab -H rake RAILS_ENV=production gitlab:backup:restore BACKUP=/home/gitlab/gitlab/tmp/backups/20130125_11h35_1359131740
Note: Backup folder is set in config/gitlab.yml. GitLab backup and restore is documented here.

Update Gitlab

When a new version is out follow the instructions at Github wiki. A new release is out every 22nd of a month.

Migrate from sqlite to mysql

Get latest code as described in #Update_Gitlab. Save data.

# cd /home/gitlab/gitlab
# sudo -u gitlab bundle exec rake db:data:dump RAILS_ENV=production

Follow #Mysql instructions and then setup the database.

# sudo -u gitlab bundle exec rake db:setup RAILS_ENV=production

Finally restore old data.

# sudo -u gitlab bundle exec rake db:data:load RAILS_ENV=production

Running GitLab with rvm

To run gitlab with rvm first you have to set up an rvm:

 curl -L https://get.rvm.io | bash -s stable --ruby=1.9.3
Note: Version 1.9.3 is currently recommended to avoid some compatibility issues.

For the complete installation you will want to be the final user (e.g. git) so make sure to switch to this user and activate your rvm:

 su - git
 source "$HOME/.rvm/scripts/rvm"

Then continue with the installation instructions from above. However, the systemd scripts will not work this way, because the environment for the rvm is not activated. The recommendation here is to create to separate shell scripts for puma and sidekiq to activate the environment and then start the service:

source `/home/git/.rvm/bin/rvm 1.9.3 do rvm env --path`
RAILS_ENV=production bundle exec puma -C "/home/git/gitlab/config/puma.rb"
source `/home/git/.rvm/bin/rvm 1.9.3 do rvm env --path`
case $1 in
        bundle exec rake sidekiq:start RAILS_ENV=production
        bundle exec rake sidekiq:stop RAILS_ENV=production
        echo "Usage $0 {start|stop}"

Then modify the above systemd files so they use these scripts. Modify the given lines:

ExecStart=/home/git/bin/sidekiq.sh start
ExecStop=/home/git/bin/sidekiq.sh stop


Sometimes things may not work as expected. Be sure to visit the Trouble Shooting Guide.

See also