Message Queue Service using Kafka

In this article, we are going to create a Message Queue Service using Kafka and KafkaJS, somewhere similar to SQS, and we will make sure that messages are processed exactly once, in the exact order that they are sent. We will also implement the redelivery mechanism to make sure that if something fails in our business logic, we can add it to the queue again.

Find the source code on Github: kafka-message-queue

Prerequisites

1. Get Kafka

Download Kafka and extract it from here, and navigate to the directory.

$ tar -xzf <file_name>.tgz
$ cd <file_name>

2. Start the Kafka Environment

Your local environment must have Java 8+ installed.

Run the following commands to start ZooKeeper:

# Start the ZooKeeper service
# Note: Soon, ZooKeeper will no longer be required by Apache Kafka.
$ bin/zookeeper-server-start.sh config/zookeeper.properties

Open another terminal session and run Kafka Broker service using:

# Start the Kafka broker service
$ bin/kafka-server-start.sh config/server.properties

Once all services have successfully launched, you will have a basic Kafka environment running and ready to use.

3. Create a Topic to Store our Queue Messages

With the following command, we will create a topic called "message-queue" in our Kafka.

$ bin/kafka-topics.sh --create --topic message-queue --bootstrap-server localhost:9092

To check the created topic use this command.

$ bin/kafka-topics.sh --describe --topic message-queue --bootstrap-server localhost:9092

You can create a topic with any number of partitions. In here we are using 1 partition only.

Creating Services Using KafkaJS

We basically need 3 things to make our queue system work perfectly.

Sending Events to Kafka Queue topic.
Subscribing to the topic and reading the Queue Message.
Handling the redelivery.

kafka-config.js

The following snippet contains the basic config of Kafka that we need for our queue system.

// Using KafkaJs nodejs library
import { Kafka } from 'kafkajs';

// kafka broker running on localhost:9092 default port
const kafkaBroker = 'localhost:9092';

// kafka topic used for queue messages
export const kafkaTopic = 'message-queue';

// kafka client with basic config
export const KafkaClient = new Kafka({
  brokers: [kafkaBroker]
});

producer.js

This service will send messages to the Kafka topic we created earlier. We can use sendMessageToQueue function and pass the message object which needs to be sent to Kafka.

import { KafkaClient } from './index.js';
import { kafkaTopic } from "./kafka-config";

export const sendMessageToQueue = async (message) => {
  const producer = KafkaClient.producer();
  await producer.connect();
  await producer.send({
    topic: kafkaTopic,
    messages: [
      {
        value: message
      }
    ]
  });
  // Disconnect producer once message sending is done.
  await producer.disconnect();
};

message-queue.js

This is our main service that takes care of receiving the messages from the Kafka queue, doing the business logic, and handling the redelivery if something goes wrong in business logic.

import { sendMessageToQueue } from "./producer.js";
import { KafkaClient, kafkaTopic, kafkaGroupId } from "./kafka-config.js";

export const consumeMessage = async () => {
  // Creating a Consumer Instance
  const consumer = KafkaClient.consumer({
    groupId: kafkaGroupId,
  });

  await consumer.connect();
  // Subscribing to our Kafka topic
  await consumer.subscribe({ topic: kafkaTopic, fromBeginning: true });

  await consumer.run({
    autoCommit: false,
    eachMessage: async ({ topic, partition, message }) => {
      const messageData = message.value.toString();
      try {
        // Do the business Logic
        console.info('Received Message', messageData);
      } catch (error) {
        console.error(error);
        // Resending message to kafka queue for redelivery
        await sendMessageToQueue(messageData);
      } finally {
        const offset = +message.offset + 1;
        // Committing the message offset to Kafka
        await consumer.commitOffsets([
          { topic: kafkaTopic, partition, offset: offset.toString() }
        ]);
      }
    }
  });
};

Setting autoCommit: false ensures the message won't be acknowledged to Kafka until we do it manually. This is critical — if our business logic throws an error, we catch it and resend the message back to the queue for redelivery. Only after successful processing (or re-queuing on failure) do we commit the offset.

Download the source: kafka-message-queue Bonus: It contains test sample files too.

References:

#Prerequisites

#1. Get Kafka

#2. Start the Kafka Environment

#3. Create a Topic to Store our Queue Messages

#Creating Services Using KafkaJS

#kafka-config.js

#producer.js

#message-queue.js