Installing and Running Kafka on an AWS Instance

In this blog, we will install and start a single-node, latest and recommended version of Kafka 0.10.2.0 with the binary for Scala 2.12 on the EC2 Linux instance with centOS as its operating system. We will be using the t2.micro (free tier) instance which comes with 1 GB RAM and 8 GB SSD.

Prerequisites

1) Create an EC2 Instance

Steps for creating an AWS instance are clearly mentioned in the official AWS documentation; check here.

2) Install Java 8

Since we will be working with the Kafka binary for Scala 2.12, our instance must have Java 8. By default, the EC2 instances have Java 7. You may check and upgrade the Java version to 8 on your instance by following the steps here.

After installing Java 8, follow the steps mentioned below in sequence to start Kafka on your instance.

Step 1: Downloading and Extracting Kafka

Download kafka_2.12-0.10.2.0.tgz

wget http://mirror.fibergrid.in/apache/kafka/0.10.2.0/kafka_2.12-0.10.2.0.tgz

Extract the .tgz file

tar -xzf kafka_2.12-0.10.2.0.tgz

Since the zip is of no use now, we remove it:

rm kafka_2.12-0.10.2.0.tgz

Step 2: Starting Zookeeper

Since Kafka uses Zookeeper, we need to first start a Zookeeper server. We can use the convenience script packaged with Kafka to start a single-node Zookeeper instance or we can start Zookeeper on a standalone instance and specify its configurations in zookeeper.properties configuration file, we would be starting it using the convenience script that is packaged with Kafka. Since we have 1 GB RAM we would be setting KAFKA_HEAP_OPTS environment variable in our .bashrc to 50% of total RAM ie 500 MB in our case.

vi .bashrc

Insert the following environment variable.

export KAFKA_HEAP_OPTS="-Xmx500M -Xms500M"

After setting the variable, source your .baschrc.

source .bashrc

Start Zookeeper by the following command in background using nohup and divert its logs in zookeeper-logs file.

cd kafka_2.12-0.10.2.0
nohup bin/zookeeper-server-start.sh config/zookeeper.properties > ~/zookeeper-logs &

Then press ctrl+d to log out of the instance.

Ssh to your instance again and check the content of zookeeper-logs file. It must look like this:

NOTE: In case the content of zookeeper-logs file is different, try freeing the RAM buffers/cache by following commands and re-run the cd and nohup command mentioned above (this may happen as t2.micro instance comes with quite less RAM, unlikely to happen with bigger instances).

sudo sh -c 'echo 1 >/proc/sys/vm/drop_caches'
sudo sh -c 'echo 2 >/proc/sys/vm/drop_caches'
sudo sh -c 'echo 3 >/proc/sys/vm/drop_caches'

Step 3: Starting Kafka

After successfully staring Zookeeper, it's now time to start Kafka via the following command:

cd kafka_2.12-0.10.2.0
nohup bin/kafka-server-start.sh config/server.properties > ~/kafka-logs &

Then press ctrl+d to log out of the instance.

Ssh to your instance again and check the content of Kafka-logs file. It must look like this:

This successfully starts Kafka on your EC2 instance. You may access your Kafka-server via Kafka-Scala or Kafka-Java API by making the required changes in the security groups. To stop Kafka and Zookeeper, enter the following commands:

bin/kafka-server-stop.sh
bin/zookeeper-server-stop.sh

Hope this blog helps you. Comments and suggestions are welcomed.