MusicBrainz Server/Setup

From MusicBrainz Wiki
< MusicBrainz Server
Revision as of 09:58, 30 September 2013 by JonnyJD (talk | contribs) (looks like the extracted/imported drive is always 27 GB right from the start. (reported by Navap for VMware))
Jump to navigationJump to search

MusicBrainz Server virtual machine

Running a MusicBrainz Server as a virtual machine requires some Linux knowledge, but it is vastly simpler than installing the server from scratch. The pre-built virtual image can be imported into either VirtualBox or VMware. If you are using Amazon EC2 you can not use this virtual image and will instead have to follow the steps outlined in the source code INSTALL.md. To set up a virtual machine instance, download the torrent file and follow these steps:

MusicBrainz Server Virtual Machine
Download via BitTorrent: MusicBrainz VM 2013-08-01
Size: 10.13GB Open Virtualization Archive (OVA)
(the extracted virtual drive is then 27 GB and can extend up to 40 GB)
Version: 2013-08-01 (beta version)
MD5: 717726cbea28a4fce51ae6ae803b8a6a

Running with VMware

  1. Download VMware Player for Windows/Linux or VMware Fusion for Mac.
  2. Import the downloaded .ova

Running with VirtualBox

Currently (20130801 beta VM) the .ova file doesn't work directly with VirtualBox so you have to do some additional steps until a fixed VM is realeased.

  1. Download and install VirtualBox on your machine.
  2. Extract the virtual disk (vmdk) from the ova file and create a VirtualBox hard disk (vdi). This can be done with:
    tar -xf "MusicBrainz 2013-08-01.ova" && VBoxManage clonehd MusicBrainz_2013-08-01-disk1.vmdk MusicBrainz130801.vdi --format VDI
    This command will create a 27 GB file (the virtual hard drive), which can grow up to 40 GB later on when using the VM.
    Make sure you don't use the vmdk directly after extracting it, since the VBoxManage command might fail otherwise. If that happens you should untar the .ova again to get an untouched vmdk.
  3. Create a new VirtualBox VM with Type "Linux" and Version "Ubuntu (64 bit)". You should give at least 1 GB RAM.
  4. Add the previously generated MusicBrainz130801.vdi as hard disk

Running with QEMU/KVM

  1. Extract the disk image:
    tar -xf MusicBrainz\ 2013-08-01.ova
  2. Covert it from VMWare format:
    virt-convert -D qcow2 MusicBrainz\ 2013-08-01.ovf /var/lib/libvirt/images
  3. Create and start the new VM:
    virt-image --os-type=linux --os-variant=ubuntuprecise /var/lib/libvirt/images/Musicbrainz-Server-2013-08-01.virt-image.xml
    This only works on a 64-bit host machine. If you wish to try it on a 32-bit host use the following instead, but beware that performance will be frustratingly poor:
    virt-install --virt-type qemu --os-type=linux --os-variant=ubuntuprecise --arch x86_64 --ram 1024 --name MusicBrainz_2013-08-01 --import --disk /var/lib/libvirt/images/MusicBrainz_2013-08-01-disk1.qcow2

Starting the VM

  1. Once the instance has started up, log in on the console using the username vm and password musicbrainz. This account has sudo privileges -- if you would like to set a root password, you can do that via sudo.
  2. Get the IP address of your virtual machine. Note this down from "inet addr" with this command:
    ifconfig | grep eth0 -A 1
    NOTE: if eth0 is not configured correctly, perhaps you've encountered this problem.
  3. Optional: The console for Virtual Box is very slow. It may be faster to SSH into the virtual box with a good terminal program.
  4. The MusicBrainz server starts automatically. Now you can reach the MusicBrainz server by pointing your browser to port 5000 of the IP address in step 5.
    • e.g If your IP address from step 6 was: 10.1.1.104, then point your browser to http://10.1.1.104:5000
    • If the server looks like it started correctly and you cannot connect in your browser, try pinging it from your host operating system command line
      ping <inet addr>
      If you cannot ping it, you may have a VirtualBox network configuration problem. Try the instructions here to switch your VM to using a bridged connection rather than NAT. Restart the virtual machine and find the new IP address in step 6 again.

Tuning your VM

We recommend that you give your VM 2GB of ram, if that is possible. The more RAM you give to the VM, the faster it is going to run. To change the memory settings, you will need to shut down the VM, change the settings and then re-start the VM. The Postgres database will be automatically tuned based on the available RAM when the VM starts up.

Running Replication

This VM comes "replication ready". To enable replication, and have the database catch up with the latest replication packets, do this:

bin/replicate now

This will load all of the changes to the database since the VM update.

NOTE: Loading replication changes might take a long time. If the VM is more than a couple of weeks old, it might be better for you to import a fresh data set. Drop pre-filled database using command

dropdb musicbrainz_db

and check the INSTALL.md file for how to import new data.

Automating Replication

To turn on background replication, run:

bin/replicate start

to turn it off:

bin/replicate stop

We recommend leaving replication off for the time being, until you've built search indexes for the VM.

Building search indexes

The VM comes with support to build search indexes. In order to build the indexes, log in to the account and then:

bin/reindex

Depending on your machine, this may take quite a long time. We recommend that you leave this running overnight. After the indexes are complete, you should be able to carry out indexed searches in your VM.

Accessing the database

To access the main postgres database, you can do this:

sudo su - musicbrainz
cd musicbrainz-server/admin
./psql READWRITE

If you would like to access the DB from outside the virtual box, take a look at how to change postgres connection settings.

Turning the VM into development box

If you would like to use the VM to do development instead of using it as a simple database slave, you'll need to edit lib/DBDefs.pm and set REPLICATION_TYPE to RT_STANDALONE and run admin/psql READWRITE and execute the following queries:

DELETE FROM annotation WHERE editor > (SELECT max(id) FROM editor);
DELETE FROM release_annotation WHERE NOT EXISTS (SELECT 1 FROM annotation WHERE annotation.id = release_annotation.annotation);

then from the command line execute:

admin/psql READWRITE < admin/sql/CreateFKConstraints.sql
admin/psql READWRITE < admin/sql/CreateFunctions.sql

TODO: The server will probably run out of disk space during this process. We need to add instructions on how to move the DB to a new partition.

Setup MusicBrainz Server from source code

This can potentially be a very laborious and time consuming method of getting a functioning MusicBrainz server. Using the virtual machine is recommended.

Get a copy of musicbrainz-server from git:

git clone --recursive https://github.com/metabrainz/musicbrainz-server.git musicbrainz-server
cd musicbrainz-server

And follow the instructions in the INSTALL file.

Support

The setup process may look daunting, but please don't let this discourage you; the INSTALL is thorough and contains a lot of information, and we are willing to provide assistance. If you have questions about installing, join us in the #musicbrainz-devel IRC channel or post a question on the developers mailing list and we will attempt to help you out.

We recommend that you dive in and give it a try - who knows how far you'll get and what you might learn along the way!

Requirements

In order to set up a running MusicBrainz server with the full database you will need:

  • A linux box, preferably Ubuntu.
  • 8GB of free disk space, (if you are a developer and only want the server code and database structure 2GB is more than enough).
  • Git knowledge which will enable you to check out the source code.

As a developer the following knowledge/skills are beneficial:

  • Apache, Perl, mod_perl, PostgreSQL and a number of perl modules.
  • How to compile and install packages from source on a Linux box.
  • How to patch existing packages, although we can help you out if you have questions about that.

Note: The server has never been ported to Windows, and we suspect that it would be a fair amount of work to make that happen.