Thooms / kafka-node

Node.js client for Apache Kafka 0.8

Kafka-node

Kafka-node is a Node.js client with Zookeeper integration for Apache Kafka 0.8.1 and later.

Table of Contents

• Features
• Install Kafka
• API
  • Client
  • Producer
  • HighLevelProducer
  • Consumer
  • HighLevelConsumer
  • ConsumerGroup
  • Offset
• Troubleshooting / FAQ
  • HighLevelProducer with KeyedPartitioner errors on first send
  • How do I debug an issue?
  • For a new consumer how do I start consuming from the latest message in a partition?
  • FailedToRebalanceConsumerError: Exception: NODE_EXISTS[-110]
  • HighLevelConsumer does not consume on all partitions
  • How to throttle messages / control the concurrency of processing messages
  • How do I consume binary data?
• Running Tests
• LICENSE - "MIT"

Features

• Consumer and High Level Consumer
• Producer and High Level Producer
• Manage topic Offsets
• SSL connections to brokers (Kafka 0.9+)
• Consumer Groups managed by Kafka coordinator (Kafka 0.9+)

Install Kafka

Follow the instructions on the Kafka wiki to build Kafka 0.8 and get a test broker up and running.

API

Client

Client(connectionString, clientId, [zkOptions], [noAckBatchOptions], [sslOptions])

• connectionString: Zookeeper connection string, default localhost:2181/
• clientId: This is a user-supplied identifier for the client application, default kafka-node-client
• zkOptions: Object, Zookeeper options, see node-zookeeper-client
• noAckBatchOptions: Object, when requireAcks is disabled on Producer side we can define the batch properties, 'noAckBatchSize' in bytes and 'noAckBatchAge' in milliseconds. The default value is { noAckBatchSize: null, noAckBatchAge: null } and it acts as if there was no batch
• sslOptions: Object, options to be passed to the tls broker sockets, ex. { rejectUnauthorized: false } (Kafka +0.9)

close(cb)

Closes the connection to Zookeeper and the brokers so that the node process can exit gracefully.

• cb: Function, the callback

Producer

Producer(client, [options])

• client: client which keeps a connection with the Kafka server.
• options: options for producer,

{
    // Configuration for when to consider a message as acknowledged, default 1
    requireAcks: 1,
    // The amount of time in milliseconds to wait for all acks before considered, default 100ms
    ackTimeoutMs: 100,
    // Partitioner type (default = 0, random = 1, cyclic = 2, keyed = 3), default 0
    partitionerType: 2
}

var kafka = require('kafka-node'),
    Producer = kafka.Producer,
    client = new kafka.Client(),
    producer = new Producer(client);

Events

• ready: this event is emitted when producer is ready to send messages.
• error: this is the error event propagates from internal client, producer should always listen it.

send(payloads, cb)

• payloads: Array,array of ProduceRequest, ProduceRequest is a JSON object like:

{
   topic: 'topicName',
   messages: ['message body'], // multi messages should be a array, single message can be just a string or a KeyedMessage instance
   key: 'theKey', // only needed when using keyed partitioner
   partition: 0, // default 0
   attributes: 2 // default: 0
}

• cb: Function, the callback

attributes controls compression of the message set. It supports the following values:

• 0: No compression
• 1: Compress using GZip
• 2: Compress using snappy

Example:

var kafka = require('kafka-node'),
    Producer = kafka.Producer,
    KeyedMessage = kafka.KeyedMessage,
    client = new kafka.Client(),
    producer = new Producer(client),
    km = new KeyedMessage('key', 'message'),
    payloads = [
        { topic: 'topic1', messages: 'hi', partition: 0 },
        { topic: 'topic2', messages: ['hello', 'world', km] }
    ];
producer.on('ready', function () {
    producer.send(payloads, function (err, data) {
        console.log(data);
    });
});

producer.on('error', function (err) {})

⚠️WARNING: Batch multiple messages of the same topic/partition together as an array on the messages attribute otherwise you may lose messages!

createTopics(topics, async, cb)

This method is used to create topics on the Kafka server. It only works when auto.create.topics.enable, on the Kafka server, is set to true. Our client simply sends a metadata request to the server which will auto create topics. When async is set to false, this method does not return until all topics are created, otherwise it returns immediately.

• topics: Array, array of topics
• async: Boolean, async or sync
• cb: Function, the callback

Example:

var kafka = require('kafka-node'),
    Producer = kafka.Producer,
    client = new kafka.Client(),
    producer = new Producer(client);
// Create topics sync
producer.createTopics(['t','t1'], false, function (err, data) {
    console.log(data);
});
// Create topics async
producer.createTopics(['t'], true, function (err, data) {});
producer.createTopics(['t'], function (err, data) {});// Simply omit 2nd arg

HighLevelProducer

HighLevelProducer(client, [options])

• client: client which keeps a connection with the Kafka server. Round-robins produce requests to the available topic partitions
• options: options for producer,

{
    // Configuration for when to consider a message as acknowledged, default 1
    requireAcks: 1,
    // The amount of time in milliseconds to wait for all acks before considered, default 100ms
    ackTimeoutMs: 100,
    // Partitioner type (default = 0, random = 1, cyclic = 2, keyed = 3), default 2
    partitionerType: 3
}

var kafka = require('kafka-node'),
    HighLevelProducer = kafka.HighLevelProducer,
    client = new kafka.Client(),
    producer = new HighLevelProducer(client);

Events

• ready: this event is emitted when producer is ready to send messages.
• error: this is the error event propagates from internal client, producer should always listen it.

send(payloads, cb)

• payloads: Array,array of ProduceRequest, ProduceRequest is a JSON object like:

{
   topic: 'topicName',
   messages: ['message body'], // multi messages should be a array, single message can be just a string,
   key: 'theKey', // only needed when using keyed partitioner
   attributes: 1
}

• cb: Function, the callback

Example:

var kafka = require('kafka-node'),
    HighLevelProducer = kafka.HighLevelProducer,
    client = new kafka.Client(),
    producer = new HighLevelProducer(client),
    payloads = [
        { topic: 'topic1', messages: 'hi' },
        { topic: 'topic2', messages: ['hello', 'world'] }
    ];
producer.on('ready', function () {
    producer.send(payloads, function (err, data) {
        console.log(data);
    });
});

⚠️WARNING: Batch multiple messages of the same topic/partition together as an array on the messages attribute otherwise you may lose messages!

createTopics(topics, async, cb)

This method is used to create topics on the Kafka server. It only work when auto.create.topics.enable, on the Kafka server, is set to true. Our client simply sends a metadata request to the server which will auto create topics. When async is set to false, this method does not return until all topics are created, otherwise it returns immediately.

• topics: Array,array of topics
• async: Boolean,async or sync
• cb: Function,the callback

Example:

var kafka = require('kafka-node'),
    HighLevelProducer = kafka.HighLevelProducer,
    client = new kafka.Client(),
    producer = new HighLevelProducer(client);
// Create topics sync
producer.createTopics(['t','t1'], false, function (err, data) {
    console.log(data);
});
// Create topics async
producer.createTopics(['t'], true, function (err, data) {});
producer.createTopics(['t'], function (err, data) {});// Simply omit 2nd arg

Consumer

Consumer(client, payloads, options)

• client: client which keeps a connection with the Kafka server. Note: it's recommend that create new client for different consumers.
• payloads: Array,array of FetchRequest, FetchRequest is a JSON object like:

{
   topic: 'topicName',
   offset: 0, //default 0
}

• options: options for consumer,

{
    groupId: 'kafka-node-group',//consumer group id, default `kafka-node-group`
    // Auto commit config
    autoCommit: true,
    autoCommitIntervalMs: 5000,
    // The max wait time is the maximum amount of time in milliseconds to block waiting if insufficient data is available at the time the request is issued, default 100ms
    fetchMaxWaitMs: 100,
    // This is the minimum number of bytes of messages that must be available to give a response, default 1 byte
    fetchMinBytes: 1,
    // The maximum bytes to include in the message set for this partition. This helps bound the size of the response.
    fetchMaxBytes: 1024 * 1024,
    // If set true, consumer will fetch message from the given offset in the payloads
    fromOffset: false,
    // If set to 'buffer', values will be returned as raw buffer objects.
    encoding: 'utf8'
}

Example:

var kafka = require('kafka-node'),
    Consumer = kafka.Consumer,
    client = new kafka.Client(),
    consumer = new Consumer(
        client,
        [
            { topic: 't', partition: 0 }, { topic: 't1', partition: 1 }
        ],
        {
            autoCommit: false
        }
    );

on('message', onMessage);

By default, we will consume messages from the last committed offset of the current group

• onMessage: Function, callback when new message comes

Example:

consumer.on('message', function (message) {
    console.log(message);
});

on('error', function (err) {})

on('offsetOutOfRange', function (err) {})

addTopics(topics, cb, fromOffset)

Add topics to current consumer, if any topic to be added not exists, return error

• topics: Array, array of topics to add
• cb: Function,the callback
• fromOffset: Boolean, if true, the consumer will fetch message from the specified offset, otherwise it will fetch message from the last commited offset of the topic.

Example:

consumer.addTopics(['t1', 't2'], function (err, added) {
});

or

consumer.addTopics([{ topic: 't1', offset: 10 }], function (err, added) {
}, true);

removeTopics(topics, cb)

• topics: Array, array of topics to remove
• cb: Function, the callback

Example:

consumer.removeTopics(['t1', 't2'], function (err, removed) {
});

commit(cb)

Commit offset of the current topics manually, this method should be called when a consumer leaves

• cb: Function, the callback

Example:

consumer.commit(function(err, data) {
});

setOffset(topic, partition, offset)

Set offset of the given topic

• topic: String

• partition: Number

• offset: Number

Example:

consumer.setOffset('topic', 0, 0);

pause()

Pause the consumer. Calling pause does not automatically stop messages from being emitted. This is because pause just stops the kafka consumer fetch loop. Each iteration of the fetch loop can obtain a batch of messages (limited by fetchMaxBytes).

resume()

Resume the consumer. Resumes the fetch loop.

pauseTopics(topics)

Pause specify topics

consumer.pauseTopics([
    'topic1',
    { topic: 'topic2', partition: 0 }
]);

resumeTopics(topics)

Resume specify topics

consumer.resumeTopics([
    'topic1',
    { topic: 'topic2', partition: 0 }
]);

close(force, cb)

• force: Boolean, if set to true, it forces the consumer to commit the current offset before closing, default false

Example

consumer.close(true, cb);
consumer.close(cb); //force is disabled

HighLevelConsumer

HighLevelConsumer(client, payloads, options)

• client: client which keeps a connection with the Kafka server.
• payloads: Array,array of FetchRequest, FetchRequest is a JSON object like:

{
   topic: 'topicName'
}

• options: options for consumer,

{
    // Consumer group id, default `kafka-node-group`
    groupId: 'kafka-node-group',
    // Optional consumer id, defaults to groupId + uuid
    id: 'my-consumer-id',
    // Auto commit config
    autoCommit: true,
    autoCommitIntervalMs: 5000,
    // The max wait time is the maximum amount of time in milliseconds to block waiting if insufficient data is available at the time the request is issued, default 100ms
    fetchMaxWaitMs: 100,
    // This is the minimum number of bytes of messages that must be available to give a response, default 1 byte
    fetchMinBytes: 1,
    // The maximum bytes to include in the message set for this partition. This helps bound the size of the response.
    fetchMaxBytes: 1024 * 1024,
    // If set true, consumer will fetch message from the given offset in the payloads
    fromOffset: false,
    // If set to 'buffer', values will be returned as raw buffer objects.
    encoding: 'utf8'
}

Example:

var kafka = require('kafka-node'),
    HighLevelConsumer = kafka.HighLevelConsumer,
    client = new kafka.Client(),
    consumer = new HighLevelConsumer(
        client,
        [
            { topic: 't' }, { topic: 't1' }
        ],
        {
            groupId: 'my-group'
        }
    );

on('message', onMessage);

By default, we will consume messages from the last committed offset of the current group

• onMessage: Function, callback when new message comes

Example:

consumer.on('message', function (message) {
    console.log(message);
});

on('error', function (err) {})

on('offsetOutOfRange', function (err) {})

addTopics(topics, cb, fromOffset)

Add topics to current consumer, if any topic to be added not exists, return error

• topics: Array, array of topics to add
• cb: Function,the callback
• fromOffset: Boolean, if true, the consumer will fetch message from the specified offset, otherwise it will fetch message from the last commited offset of the topic.

Example:

consumer.addTopics(['t1', 't2'], function (err, added) {
});

or

consumer.addTopics([{ topic: 't1', offset: 10 }], function (err, added) {
}, true);

removeTopics(topics, cb)

• topics: Array, array of topics to remove
• cb: Function, the callback

Example:

consumer.removeTopics(['t1', 't2'], function (err, removed) {
});

commit(cb)

Commit offset of the current topics manually, this method should be called when a consumer leaves

• cb: Function, the callback

Example:

consumer.commit(function(err, data) {
});

setOffset(topic, partition, offset)

Set offset of the given topic

• topic: String

• partition: Number

• offset: Number

Example:

consumer.setOffset('topic', 0, 0);

pause()

Pause the consumer. Calling pause does not automatically stop messages from being emitted. This is because pause just stops the kafka consumer fetch loop. Each iteration of the fetch loop can obtain a batch of messages (limited by fetchMaxBytes).

resume()

Resume the consumer. Resumes the fetch loop.

close(force, cb)

• force: Boolean, if set to true, it forces the consumer to commit the current offset before closing, default false

Example:

consumer.close(true, cb);
consumer.close(cb); //force is disabled

ConsumerGroup

The new consumer group uses Kafka broker coordinators instead of Zookeeper to manage consumer groups. This is supported in Kafka version 0.9 and above only.

Coming from the highLevelConsumer

API is very similar to HighLevelConsumer with some exceptions noted below:

• In an effort to make the API simpler you no longer need to create a client this is done inside the ConsumerGroup
• consumer ID do not need to be defined. There's a new ID to represent consumers called member ID and this is assigned to consumer after joining the group
• Offsets, group members, and ownership details are not stored in Zookeeper
• ConsumerGroup does not emit a registered event

ConsumerGroup(options, topics)

var options = {
  host: 'zookeeper:2181',
  zk : undefined,   // put client zk settings if you need them (see Client)
  batch: undefined, // put client batch settings if you need them (see Client)
  ssl: true, // optional (defaults to false) or tls options hash
  groupId: 'ExampleTestGroup',
  sessionTimeout: 15000,
  // An array of partition assignment protocols ordered by preference.
  // 'roundrobin' or 'range' string for built ins (see below to pass in custom assignment protocol) 
  protocol: ['roundrobin'],
  
  // Offsets to use for new groups other options could be 'earliest' or 'none' (none will emit an error if no offsets were saved)
  // equivalent to Java client's auto.offset.reset
  fromOffset: 'latest', // default
  
  migrateHLC: false,    // for details please see Migration section below
  migrateRolling: true
};

var consumerGroup = new ConsumerGroup(options, ['RebalanceTopic', 'RebalanceTest']);

// Or for a single topic pass in a string

var consumerGroup = new ConsumerGroup(options, 'RebalanceTopic');

Custom Partition Assignment Protocol

You can pass a custom assignment strategy to the protocol array with the interface:

string :: name

integer :: version

object :: userData

function :: assign (topicPartition, groupMembers, callback)

topicPartition

{
  "RebalanceTopic": [
    "0",
    "1",
    "2"
  ],
  "RebalanceTest": [
    "0",
    "1",
    "2"
  ]
}

groupMembers

[
  {
    "subscription": [
      "RebalanceTopic",
      "RebalanceTest"
    ],
    "version": 0,
    "id": "consumer1-8db1b117-61c6-4f91-867d-20ccd1ad8b3d"
  },
  {
    "subscription": [
      "RebalanceTopic",
      "RebalanceTest"
    ],
    "version": 0,
    "id": "consumer3-bf2d11f4-1c73-4a39-b498-cfe76eb65bea"
  },
  {
    "subscription": [
      "RebalanceTopic",
      "RebalanceTest"
    ],
    "version": 0,
    "id": "consumer2-9781058e-fad4-40e8-a69c-69afbae05184"
  }
]

callback(error, result)

result

[
  {
    "memberId": "consumer3-bf2d11f4-1c73-4a39-b498-cfe76eb65bea",
    "topicPartitions": {
      "RebalanceTopic": [
        "2"
      ],
      "RebalanceTest": [
        "2"
      ]
    },
    "version": 0
  },
  {
    "memberId": "consumer2-9781058e-fad4-40e8-a69c-69afbae05184",
    "topicPartitions": {
      "RebalanceTopic": [
        "1"
      ],
      "RebalanceTest": [
        "1"
      ]
    },
    "version": 0
  },
  {
    "memberId": "consumer1-8db1b117-61c6-4f91-867d-20ccd1ad8b3d",
    "topicPartitions": {
      "RebalanceTopic": [
        "0"
      ],
      "RebalanceTest": [
        "0"
      ]
    },
    "version": 0
  }
]

Auto migration from the v0.8 based highLevelConsumer

We have two options for automatic migration from existing highLevelConsumer group. This is useful to preserve the previous committed offsets for your group.

We support two use cases:

1. You have downtime and your old HLC consumers are no longer available
2. Where the old HLC group is still up and working and you are doing a rolling deploy with zero downtime

For case 1 use below setting:

{
	migrateHLC: true, // default is false
	migrateRolling: false // default is true
}

For case 2 setting migrateRolling to true will allow the ConsumerGroup to start monitoring zk nodes for when topic ownership are relinquished by the old HLC consumer. Once this is done the ConsumerGroup will connect and the previous HLC offsets from zookeeper will be migrated automatically to the new Kafka broker based coordinator.

• Group name should be consistent with old highLevelConsumer
• Should never overwrite existing offsets
• Only offsets for Topics that were once in the highLevelConsumer will be migrated over offsets for new topics will follow the fromOffset setting

Offset

Offset(client)

• client: client which keeps a connection with the Kafka server.

events

• ready: when zookeeper is ready
• connect when broker is ready

fetch(payloads, cb)

Fetch the available offset of a specific topic-partition

• payloads: Array,array of OffsetRequest, OffsetRequest is a JSON object like:

{
   topic: 'topicName',
   partition: 0, //default 0
   // time:
   // Used to ask for all messages before a certain time (ms), default Date.now(),
   // Specify -1 to receive the latest offsets and -2 to receive the earliest available offset.
   time: Date.now(),
   maxNum: 1 //default 1
}

• cb: Function, the callback

Example

var kafka = require('kafka-node'),
    client = new kafka.Client(),
    offset = new kafka.Offset(client);
    offset.fetch([
        { topic: 't', partition: 0, time: Date.now(), maxNum: 1 }
    ], function (err, data) {
        // data
        // { 't': { '0': [999] } }
    });

commit(groupId, payloads, cb)

• groupId: consumer group
• payloads: Array,array of OffsetCommitRequest, OffsetCommitRequest is a JSON object like:

{
   topic: 'topicName',
   partition: 0, //default 0
   offset: 1,
   metadata: 'm', //default 'm'
}

Example

var kafka = require('kafka-node'),
    client = new kafka.Client(),
    offset = new kafka.Offset(client);
    offset.commit('groupId', [
        { topic: 't', partition: 0, offset: 10 }
    ], function (err, data) {
    });

fetchCommits(groupid, payloads, cb)

Fetch the last committed offset in a topic of a specific consumer group

• groupId: consumer group
• payloads: Array,array of OffsetFetchRequest, OffsetFetchRequest is a JSON object like:

{
   topic: 'topicName',
   partition: 0 //default 0
}

Example

var kafka = require('kafka-node'),
    client = new kafka.Client(),
    offset = new kafka.Offset(client);
    offset.fetchCommits('groupId', [
        { topic: 't', partition: 0 }
    ], function (err, data) {
    });

fetchLatestOffsets(topics, cb)

Example

	var partition = 0;
	var topic = 't';
	offset.fetchLatestOffsets([topic], function (error, offsets) {
		if (error)
			return handleError(error);
		console.log(offsets[topic][partition]);
	});

Troubleshooting / FAQ

HighLevelProducer with KeyedPartitioner errors on first send

Error:

BrokerNotAvailableError: Could not find the leader

Call client.refreshMetadata() before sending the first message. Reference issue #354

How do I debug an issue?

This module uses the debug module so you can just run below before starting your app.

export DEBUG=kafka-node:*

For a new consumer how do I start consuming from the latest message in a partition?

1. Call offset.fetchLatestOffsets to get fetch the latest offset
2. Consume from returned offset

Reference issue #342

FailedToRebalanceConsumerError: Exception: NODE_EXISTS[-110]

This error can occur when a HLC is killed and restarted quickly. The ephemeral nodes linked to the previous session are not relinquished in zookeeper when SIGINT is sent and instead relinquished when zookeeper session timeout is reached. The timeout can be adjusted using the sessionTimeout zookeeper option when the Client is created (the default is 30000ms).

Example handler:

process.on('SIGINT', function () {
    highLevelConsumer.close(true, function () {
        process.exit();
    });
});

Alternatively, you to avoid this issue entirely by omitting the HLC's id and a unique one will be generated for you.

Reference issue #90

HighLevelConsumer does not consume on all partitions

Your partition will be stuck if the fetchMaxBytes is smaller than the message produced. Increase fetchMaxBytes value should resolve this issue.

Reference to issue #339

How to throttle messages / control the concurrency of processing messages

1. Create a async.queue with message processor and concurrency of one (the message processor itself is wrapped with setImmediate so it will not freeze up the event loop)
2. Set the queue.drain to resume the consumer
3. The handler for consumer's message event pauses the consumer and pushes the message to the queue.

How do I consume binary data?

In the consumer set the encoding option to buffer there not settings on producers needed for this.

Reference to issue #470

Running Tests

Install Docker

On the Mac install Docker for Mac.

Start Docker and Run Tests

npm test

Stop Docker

npm run stopDocker

LICENSE - "MIT"

Copyright (c) 2015 Sohu.com

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.