After completing this (so-called) guide, I found that it’s too long and too FreeBSD-specifics.

So I decided to write a more generic documentation. This page is generalization from previous guide. And this article is:

• Much more informative than current INSTALL
• Translated version of INSTALL.debian (with exception for mongrel etc)
• Simplified edition of my previous guide
• Much like older version of INSTALL which was rather informative but for some reason get extra-simplified to the degree not informative at all (except for pointing to INSTALL.debian)

This documentation assumes you will be using Unicorn and nginx.

Software Requirement

• GD image processing library
• PostgreSQL >= 8.3 + contrib
• memcached (optional)
• Ruby
• Ruby Gems, with following gems:
  • pg
  • daemons
  • diff-lcs
  • html5
  • memcache-client
  • json
  • rails (version 2.3.4)
  • aws-s3
  • mechanize
  • system_timer
  • sys-cpu
  • unicorn
• Subversion
• nginx
• gcc

Misc Requirement

• A domain name for danbooru (it won’t install on folder, ie. www.mydomain.com/danbooru)
• Several megabytes of RAM
• Full root control (maybe not needed but it’s much easier installing as root)
• Free time
• Etc

Installation

1. Ensure all the requirements are installed.
2. Initialize new database for Danbooru. Login to postgresql as superuser (with psql):

   # create database [database_name];
   # create user [username] with password '[password]';
   # alter database [database_name] owner to [username];

3. Create new user for Danbooru installation.
4. Fetch latest Danbooru source (svn://donmai.us/danbooru/trunk).
5. Create and edit configuration files (local_config.rb and database.yml) accordingly. Additional configuration options is at default_config.rb but any modification should be done at local_config.rb as it’ll override any parameters in default_config.rb. Don’t forget to set CONFIG["session_secret_key"].
6. Don’t forget to create folder public/data/sample if you enable sample creation.
7. Compile image resizer library at lib/danbooru_image_resizer (ruby extconf.rb && make) but do not make install it
8. Create and install initial database. Run script scripts/db-init.sh and then import initial database at db/postgres.sql (run > psql -U [username] -f db/postgres.sql -d [database_name])
9. Run db migration script, with environment setting RAILS_ENV=production, by doing rake db:migrate from Danbooru root folder.
10. Test installation by running > unicorn_rails and check if it runs fine at [server_ip]:8080 from a browser. If it runs fine you can start the daemon by configuring unicorn (port – whatever unused – higher than 2000, ip address – 127.0.0.1, etc).
11. Configure nginx to proxy to unicorn_rails. Add this server block:

    server {
      listen 80;
      server_name [domain_name];
      location ~ /.svn { deny all; }
      location / {
        root /path/to/danbooru/public;
        try_files $uri /maintenance.html @danbooru;
        client_max_body_size [maximum_upload_size]m;
        expires max;
      }
      location @danbooru {
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $http_host;
        proxy_redirect off;
        proxy_pass http://127.0.0.1:[unicorn_port];
      }
    }

12. Start nginx, ensure everything works and create the first (admin) account

Much simpler than previous guide (well, this is not a guide, this is documentation – or so I hope). The previous guide was intended mainly for my own usage anyway 😛

As usual, corrections etc are welcome.

2008-09-17: minor update (regarding session key config).

2010-01-16: update yaaay.