Skip to content

Latest commit

 

History

History
129 lines (84 loc) · 3.92 KB

HACKING-CAA.md

File metadata and controls

129 lines (84 loc) · 3.92 KB
Raw

Introduction

This file contains instructions for setting up a local version of the coverartarchive.org service. The coverartarchive.org service consists of several parts:

  1. The MusicBrainz server (this allows image uploads)
  2. CAA-indexer (updates image metadata on the archive.org servers)
  3. coverart_redirect (image redirect service)
  4. internet archive S3 storage

The following instructions assume you are running Ubuntu 12.04.

Storage

For development we need a place to store the images, obviously a development server should not upload images to the actual coverartarchive storage.

We provide a simple Plack script which emulates just enough of the internet archive S3 protocol to use it with the musicbrainz server.

It is not designed to server back those static files however, so for now you will also need to be running a regular webserver (apache, nginx, etc..).

Figure out where you want to store uploaded images and make sure the storage server (ssssss.psgi) has write permissions at that location. E.g. if you have default debian apache install you could do something like this:

$ sudo mkdir /var/www/caa
$ sudo chown user.users /var/www/caa
$ SSSSSS_STORAGE=/var/www/caa/ plackup --port 5050 -r contrib/ssssss.psgi

Note that we're specifying a port here. We need the default port (5000) for musicbrainz.

Now that you're running this script, mb_server should be able to upload images and indexes to http://localhost/caa/$BUCKET, where $BUCKET is the bucket name.

To simulate a 503 Slow Down error, run slowdown.psgi instead of ssssss.psgi:

$ plackup --port 5050 contrib/slowdown.psgi

CAA-indexer

Download the CAA-indexer and install RabbitMQ.

$ git clone git://github.com/metabrainz/CAA-indexer.git
$ sudo apt-get install rabbitmq
$ sudo /etc/init.d/rabbitmq start

You will also need to install the pg_amqp extension for PostgreSQL. For details on this, see https://github.com/omniti-labs/pg_amqp, but it can generally be described as:

$ git clone https://github.com/omniti-labs/pg_amqp.git
$ cd pg_amqp
$ sudo make install

And then editing postgresql.conf to have:

shared_preload_libraries = 'pg_amqp.so'

Restart postgresql for the changes in postgresql.conf to take effect.

Install the triggers into the database:

$ cd ../musicbrainz-server/
$ ./admin/psql READWRITE < ./admin/sql/caa/CreateMQTriggers.sql
$ cd -

Install the dependancies for the CAA-indexer and create a configuration file for the CAA-indexer itself:

$ cpanm --installdeps --notest .
$ cp config.ini.example config.ini
$ vim config.ini

FIXME: Currently there is no way to configure the server where indexes should be uploaded to. We will have to hardcode the correct URL in the source, edit lib/CoverArtArchive/IAS3Request.pm and change line 21.

-    my $uri = "$protocol://$1.s3.us.archive.org$2";
+    my $uri = "$protocol://localhost/caa/$1?file=$2";

And finally run the indexer:

$ ./caa-indexer

coverart_redirect

Download the coverart redirect service and install its dependancies:

$ git clone git://github.com/metabrainz/coverart_redirect.git
$ sudo apt-get install python-cherrypy3 python-psycopg2 python-sqlalchemy python-werkzeug

Create a configuration file:

$ cp coverart_redirect.conf.dist coverart_redirect.conf
$ vim coverart_redirect.conf

Set prefix=http://localhost/caa/ in the [s3] section (this should be the location where ssssss.psgi is storing uploaded images).

And start the server:

$ python ./coverart_redirect_server.py

MusicBrainz Server

Now that all of that is configured you can point the musicbrainz server at the upload and download urls for your local cover art archive, change the following values in lib/DBDefs.pm:

sub COVER_ART_ARCHIVE_UPLOAD_PREFIXER { shift; sprintf("//localhost:5050/%s", shift) };
sub COVER_ART_ARCHIVE_DOWNLOAD_PREFIX { "http://localhost:8080" };