Note
	The High Availability features are only available in the Neo4j Enterprise Edition.

To run the Neo4j server in high availability mode there are two things you need to do. You have to configure the server to start up the database in high availability mode and you have to configure the Neo4j database for operating in high availability mode.

Instructing the server to start the database in high availability mode is as easy as setting the org.neo4j.server.database.mode property in the server properties file (conf/neo-server.properties) to ha. The default value for this parameter is single, which will start the database in standalone mode without participating in a cluster, still giving you Online Backup.

Configuring the Neo4j database for operating in high availability mode requires specifying a few properties in conf/neo4j.properties. First you need to specify ha.machine_id, this is a positive integer id that uniquely identifies this server in the cluster.

Example: ha.machine_id = 1

Then you have to specify ha.zoo_keeper_servers, this is a comma separated list of hosts and ports for communicating with each member of the Neo4j Coordinator cluster.

For example: ha.zoo_keeper_servers = neo4j-manager-01:2180,neo4j-manager-02:2180,neo4j-manager-03:2180.

You can also, optionally, configure the ha.cluster_name. This is the name of the cluster this instance is supposed to join. Accepted characters are alphabetical, numerical, dot, dash, and underscore. This configuration is useful if you have multiple Neo4j HA clusters managed by the same Coordinator cluster.

Example: ha.cluster_name = my_neo4j_ha_cluster

4.4.1. Starting a Neo4j Coordinator

A Neo4j Coordinator cluster provides the Neo4j HA Data cluster with reliable coordination of lifecycle activities, like electing the master. Neo4j Server includes everything needed for running a Neo4j Coordinator.

Configuration of a Coordinator is specified in these files:

• conf/coord.cfg - coordinator operational settings
• data/coordinator/myid - unqiue identification of the coordinator

Once a Neo4j Coordinator instance has been configured, you can use the bin/neo4j-coordinator command to start the Neo4j Coordinator server on all desired servers with the same configuration, just changing the data/coordinator/myid to unique numbers. You can check that the coordinator is up by running jconsole , attaching to the JVM and check for org.apache.zookeeper MBeans.

Figure 4.1. Neo4j Coordinator MBeans View

Tue Apr 12 09:25:58 CEST 2011: MasterServer communication server started and bound to 6361
Tue Apr 12 09:25:58 CEST 2011: Started as master
Tue Apr 12 09:25:58 CEST 2011: master-rebound set to 1