Getting Started with Your First Topology - Real-time Analytics with Storm and Cassandra - page 34

Set up Zookeeper (V 3.3.5) for Storm
The coordination of a Storm topology is maintained by a Zookeeper cluster. The utilization
of Zookeeper is not very high, as it just maintains the runnable state of the Storm cluster. In
most cases, a single Zookeeper node should suffice, but in production scenarios, at least a
three-node Zookeeper cluster is recommended so that a single node doesn't become a single
point of failure.
For reliable Zookeeper service, deploy Zookeeper in a cluster known as an ensemble . As
long as the majority of the ensemble is up, the service will be available. One of the nodes in
the ensemble is automatically selected as a leader and others as followers. If the leader goes
down, one of the follower nodes becomes the leader.
Perform the following steps on all the machines that will be part of the Zookeeper en-
semble to set up the Zookeeper cluster:
1. Download the most recent stable release (version 3.3.5) from the Apache Zookeep-
er site.
2. Create a zookeeper directory under /usr/local :
sudo mkdir /usr/local/zookeeper
3. Extract the downloaded TAR file to the /usr/local location. Use the following
command:
sudo tar -xvf zookeeper-3.3.5.tar.gz -C /usr/local/
zookeeper
4. Zookeeper needs a directory to store its data. Create /usr/local/zookeep-
er/tmp to store this data:
sudo mkdir -p /usr/local/zookeeper/tmp
5. Create a configuration file, zoo.cfg , under /usr/local/zookeeper/zoo-
keeper-3.3.5/conf . The following properties will go in it:
◦ tickTime : This is the number of milliseconds of each tick (for example,
2000).
◦ initLimit : This is the number of ticks that the initial synchronization
phase can take (for example, 5).