This section contains information about how to install a Leela cluster. Is is divided into two sections. The first covers how to create the packages and the later how to install and configure the packages.

This document will use debian-wheezy as the target platform and will make use of the docker image that has been described in the environment section. If you haven't done so refer to that document to get fetch the source code and create the docker image.

Now start the container. We will use it to build every single component:

$ docker run \
    -v .:/leela \
    --rm -i -t \
    leela/debian7-{arch} /bin/bash

Also this guide will install the local packages we have just generated. You are free to use any frontend you like but as we are using local files we find it easier using gdebi.

$ apt-get install gdebi

LIBLEELA

Libleela is the client library written in C. It is the base for the other libraries written in other languages and also contain the poly1305aes library. You will need it to built other leela components.

docker $ make -C /leela/pkg libleela.debian

It is likely to fail since there might be a few dependencies missing. However those can be satisfied by the package manager. So just install those that are missing and try again.

On success the package will be available at:

docker $ ls -1 /leela/pkg/dist/debian7/{arch}/libleela/
libleela-dev_{major}.{minor}.{patch}-{build}_{arch}.deb
libleela_{major}.{minor}.{patch}-{build}.dsc
libleela_{major}.{minor}.{patch}-{build}.tar.gz
libleela_{major}.{minor}.{patch}-{build}_{arch}.changes
libleela_{major}.{minor}.{patch}-{build}_{arch}.deb

LIBLEELA-PYTHON

docker $ make -C /leela/pkg libleela-python.debian

Likewise, some dependencies may be missing. You must solve them and try again. If everything went right, you should have the following packages:

docker $ ls -1 /leela/pkg/dist/debian7/{arch}/libleela-python
libleela-python_x.y.z-b.dsc
libleela-python_x.y.z-b.tar.gz
libleela-python_x.y.z-b_{arch}.changes
libleela-python_x.y.z-b_{arch}.deb

WARPDRIVE

The warpdrive is the frontend of the Leela cluster. It is the service that allows clients to store data and retrieve information. TODO:moreinfo.

docker $ make -C /leela/pkg leela-warpdrive.debian

Which should generate the following files:

docker $ ls -1 /leela/pkg/dist/debian7/{arch}/leela-warpdrive/
leela-warpdrive_{major}.{minor}.{patch}-{build}.dsc
leela-warpdrive_{major}.{minor}.{patch}-{build}.tar.gz
leela-warpdrive_{major}.{minor}.{patch}-{build}_{arch}.changes
leela-warpdrive_{major}.{minor}.{patch}-{build}_{arch}.deb

COLLECTD

This package provides a collectd write plugin. With this you can send collectd metrics to Leela.

As usual, lets start preparing the environment. Notice that we use collectd from source, since it is a bit unusual to compile modules.

BLACKBOX

This is the storage backend used by the Leela cluster. It is used internally and it the module that gives access to the cassandra cluster. TODO:moreinfo.

It is written in clojure which means you will need java [we recommend using Oracle's version, specially in production]. We will use an Ubuntu PPA repository since debian has no java packages but openjdk.

docker $ /leela/automation/bootstrap/clojure-bootstrap.sh

After this script finishes you will have java and leiningen installed which allows us to build the package:

docker $

INSTALL

BLACKBOX

WARPDRIVE

First install the package:

$ gdebi install leela/pkg/leela-warpdrive_{major}.{minor}.{patch}-{build}_{arch}.deb

Then you must configure the daemon. The file you must change on debian is /etc/default/leela-warpdrive. It is self-documented so we just describe the most important options.

The following two variables defines the username and password used to test this warpdrive instance. Leela uses consul for auto-discovery and each service has a health checker associated. If the watchdog script fails to execute the instance does not get announced.

LEELA_WARPDRIVE_WATCHDOG_USER=
LEELA_WARPDRIVE_WATCHDOG_PASS=

Another very important configuration is:

LEELA_WARPDRIVE_ENDPOINT=

It defines network address of this instance that will be announced in consul. You probably will need to change it since it defaults to local host.

After the /etc/default/leela-warpdrive properly configured, you may start the daemon:

$ /etc/init.d/leela-warpdrive [start|stop|restart]

Notice that it depends on a number of services to work correcly. Make sure you got these dependencies satisfied before making your cluster available: