============
Installation
============

Prerequisites
=============

  * ElasticSearch 2.x
  * elasticsearch python package
  * cerbere python package for data tiling only


Installing the ElasticSearch cluster
====================================

To setup the ElasticSearch cluster, you will need to install ElasticSearch on
each node of the ES_NODES list.

This guide describes the installation of ElasticSearch from source, but you can
use your operating system package manager to install it automatically.

Requirements
------------

* SSH access to each node in ES_NODES

* Java JDK installed on each node

Instructions
------------

On one of the nodes in ES_NODES (we will call this node NODE1):

  1. Install elasticsearch

    Refer to Elasticsearch documentation for installation and set-up.

  .. important::

    Elasticsearch version must be >= 2.0.
  
  2. Install the elasticsearch-head plugin
  
  .. sourcecode:: bash
  
    cd /opt/elasticsearch-2.3
    bin/plugin -install mobz/elasticsearch-head

  3. Install naiad

  Get naiad source code:

  .. sourcecode:: bash

    git clone https://git.cersat.fr/cersat/naiad.git

  Go into the source code repository and run: 
  
  .. sourcecode:: bash
  
    sudo pip install .

  4. Other useful python packages for contribs tilers
  
  https://git.cersat.fr/cerbere/cerform


The following steps are for ElasticSearch installed on a cluster:

  5. Edit configuration file config/elasticsearch.yml
  
  .. sourcecode:: yaml
  
    cluster.name: felyx_cluster
    network.host: <IP>
    discovery.zen.minimum_master_nodes: 2
    discovery.zen.ping.multicast.enabled: false
    discovery.zen.ping.unicast.hosts: ["<IP1>", "<IP2>", ..., "<IPx>"]
  ..

  If you plan to use a cluster, but only use one node, remove/comment the
  discovery.* properties.

  * Replace <IP> by the actual IP address of the current node
  
  * Replace <IP1>, <IP2>, ..., <IPx> by the IP addresses of all the nodes
    contained in ES_NODES.
  
  5. Start ElasticSearch
  
  .. sourcecode:: bash
  
    /opt/elasticsearch-2.3/bin/elasticsearch


The first ElasticSearch node should now be up and running, all that's left is to
do the same for the other nodes.



On each remaining node:

  .. sourcecode:: bash
  
    scp -r NODE1:/opt/elasticsearch-2.3 /opt
  
  It should allow you to skip step 1., 2. and 3.
  
  Then repeat steps 4. (you should just change the "network.host" setting) and 5.

Test
====

Open a webbrowser on http://<node>:9200/_plugin/head/ (replace <node> by the IP
address of one the ElasticSearch nodes), you should see the cluster status page,
listing all the nodes currently connected.

Check that they are all listed on this page: if a node is not there, check it's
configuration and restart ElasticSearch.