Driven Administrator Guide

version 2.1.4

Installing the Driven Server

Prerequisite

Java 7 installation on the host computer

NOTES FOR UPGRADES:

If you do not want to retain Driven data after the upgrade, then perform a fresh installation of Driven Server version 2.1 as documented below.
To upgrade and migrate your data, perform an upgrade for each incremental version change between your current installation and version 2.1. See the documentation for the following upgrade paths:
- Driven 1.2 > 2.1
- Driven 1.1 > 1.2

Download and Install the Driven Server

Step 1: Download the Driven Server installation ZIP archive, which includes Apache Tomcat and embedded Elasticsearch libraries, by one of the following methods:

Download the ZIP archive directly with this hyperlink: driven-tomcat-2.1.4.zip
Alternatively, on UNIX, Linux, and Mac, run the following command:

$ wget -i http://files.concurrentinc.com/driven/2.1/driven-server/latest-tomcat.txt

The ZIP archive is packaged to address all interdependencies.

Step 2: Unzip the Driven installation package

Extract the installation package into your installation directory.

$ unzip driven-tomcat-2.1.4.zip
$ cd driven-tomcat-2.1.4

Tip	If you already have a servlet container that you want to use as the web server, deploy the WAR file from the extracted Driven installation package as the ROOT context.

Installing Elasticsearch for Clustered Production Environments

You must perform stand-alone Elasticsearch installation if you are deploying Driven on multiple nodes.

Before you begin Elasticsearch installation:

See the Driven Compatibility Matrix section of the Planning a Driven Deployment page to ensure that you install a compatible version of Elasticsearch.

It is recommended to use the operating system packages provided by elasticsearch. These will set up a user to run the elasticsearch daemon. We recommend setting up at least 3 Elasticsearch nodes for a highly available setup. We recommend also deploying on mid-size machines, since elasticsearch works better when the heap is not too big. This has to do with garbage collector settings and is beyond this guide. We recommend running elasticsearch with a heap size of 8G.

Below you will find recommended settings for a driven deployment that are different from the defaults in elasticsearch.yml. An explanation of each setting is below each item:

cluster.name: driven

Sets the name of the cluster to driven. The name can be anything, as long as the driven server and elasticsearch use the same name.

bootstrap.mlockall: true

Makes sure that the elasticsearch daemon never swaps.

indices.fielddata.cache.size: 75%

Limits the field data cache to use a maximum of 75% of the heap.

indices.memory.index_buffer_size: 50%

Allows the index buffer to take up to 50% of the heap.

threadpool.bulk.type: fixed
threadpool.bulk.size: 60
threadpool.bulk.queue_size: 300

Sets the bulk thread-pool to a fixed pool of 60 threads with a queue size of 300 items. Driven makes heavy use of the bulk api and requires a large enough queue size on busy servers.

script.disable_dynamic: false

Some functionality in Driven requires the dynamic scripting feature of Elasticsearch.