Fetcher

Your Job or Message fetching paradise, with sugar on top.

Install

Install the module using NPM:

npm install @waldo/fetcher --save

Documentation

Fetcher wraps around the node-kafka, kafka-to-sqs and sqs Waldo NPM libraries to provide a uniform and robust interface to consume and produce messages on Kafka.

The Fetcher can operate in multiple scenarios, when consuming it may consume directly from a Kafka topic or through a Kafka-to-SQS bridge if your operation is I/O or CPU intensive. On the output, you may produce a Kafka message and even choose to produce Kafka error messages in case your operation fails.

All of this is abstracted away from you, the only thing you have to provide is the initial Fetcher configuration and a Promise returning processing method.

Quick Start

const Fetcher = require(@waldo/node-fetcher);

const fetcher = new Fetcher({/* ... */});

fetcher.init()
  .then(function() {
    console.log('You are ready!');
  })
  .catch(function(err) {
    console.error('awwwOh:', err);
  });

Fetcher Configuration

When creating a new Fetcher instance you need to provide a very elaborate configuration object. All the options bellow are keys to the Object you need to pass as argument when instantiating Fetcher.

Required Config

• topic string The topic to consume kafka messages on.
• consumerGroup string The group of the consumer.
• schema Object The kafka schema of the consumer.
• process Function The master processing method, must return a promise. It will receive two arguments when invoked:
  1. uid String|null First argument will be the userId, available only if hasUser is set to true, if not it will be of value null.
  2. message Object The actual message to consume.
• log Object An object containing the methods: info, warn, error.

Optional Config

• useKeys boolean= Set to true to use Kafka Partitioning Keys.
• noKafkaSource boolean= Do not use Kafka as source, requires SQS enabled (hasSqs).

Credentials and Users

• hasUser boolean= If the consumer message has credentials with the userid encoded inside.
• credentialsKey string= The root attribute on which the credentials are stored, required if hasUser is enabled.

SQS configuration

• hasSqs boolean= Set to true if you want to activate kafka-to-sqs.
• sqsUrl string= SQS Url, required if hasSqs is enabled.
• concurrentOpsLimit number= SQS concurrent operations, required if hasSqs is enabled.

Produce to Kafka topic on response

• produceKafka boolean= Produce a kafka message upon completion using the return value of your processing method.
• topicProduce string= The topic to produce kafka messages on.
• schemaProduce Object= The kafka schema of the producer.
• keyAttribute string= Define the key attribute in the response schema to use as the kafka key for partioning.

Produce to Kafka Error topic

• produceErrorMessage boolean= Produce a kafka message in case processing fails.
• topicProduceError string= The topic to produce kafka messages on.
• schemaProduceError Object= The kafka schema of the producer.
• keyAttributeError string= Define the key attribute in the response schema to use as the kafka key for partioning.
• generateErrorMessage Function= A synchronous method that returns the correct payload to send as a kafka error message, required if produceErrorMessage is enabled, as arguments it will receive:
  1. error Error The error object.
  2. message Object The message that was attempted to be produced.

Configuration Examples

Consume and Produce to Kafka, no SQS

const opts = {
  log: log,
  topic: 'test-topic',
  consumerGroup: 'consumer-group',
  schema: avroSchemaReference,
  process: function(uid, message) {
      return Promise.resolve();
  },
  hasUser: true,
  credentialsKey: 'viewer',
  // produce related messages
  produceKafka: true,
  topicProduce: 'test-produce-topic',
  schemaProduce: schemaFix,
  keyAttribute: 'albumId',
};

const fetcher = new Fetcher(opts);

Initializing Fetcher

Once fetched is instanciated it needs to be initialized so it performs all the required connections with the services as per the configuration. The init() returns a Promise:

fetched.init()
    .then(function() {
        console.log('done');
    })

Releasing

1. Update the changelog bellow.
2. Ensure you are on master.
3. Type: grunt release
   • grunt release:minor for minor number jump.
   • grunt release:major for major number jump.

Release History

• v0.1.4, 23 Jan 2017
  • Feature: Introduce noKafkaSource option.
• v0.1.3, 20 Jan 2017
  • Upgrade kafka-to-sqs to 0.1.2.
• v0.1.2, 20 Jan 2017
  • Upgrade node-kafka to 0.1.1.
• v0.1.1, 18 Jan 2017
  • Use updated kafka-to-sqs 0.1.1.
• v0.1.0, 18 Jan 2017
  • Added the useKeys boolean option, now to use key-schemas this option needs to be set to true.
• v0.0.2, 17 Jan 2017
  • Add message as a second argument on the generateErrorMessage() callback.
• v0.0.1, 17 Jan 2017
  • Big Bang

License

Copyright Waldo, Inc. All rights reserved.