insightsdev / zookeeper-leader-election Goto Github PK

Author: Alan Perez-Rathke ([email protected])

This is a simple implementation of the ZooKeeper leader election algorithm as outlined here:

http://zookeeper.apache.org/doc/trunk/recipes.html#sc_leaderElection

Classes:

*ZooElectableClient:
- An abstract ZooKeeper client that implements the leader election algorithm.
- The method performRole() must be implemented in a child class.  This function is intended to
contain the work that the client node must actually perform (conditioned on whether or not it
is the leader).

*ZooTestElectableClient:
- A simple demo ZooKeeper client that simply outputs its state after any leader election.

*ZooZNodeDeletionMonitor: Based on http://zookeeper.apache.org/doc/current/javaExample.html#ch_Introduction
- A monitor that watches for znode deletion and informs its listener when this occurs

Scripts (*nix):

./build.sh - will compile the java sources
./run.sh - will run an instance of the test client

Usage:

The clients assume that a server is running in the background. See:

http://zookeeper.apache.org/doc/trunk/zookeeperStarted.html

for documentation on downloading and starting a ZooKeeper server.  To demo, run multiple
client instances (each in separate terminal windows), then kill the leader instance to
observe a new leader election.