Prozess

Prozess is a Kafka library for node.js

Kafka is a persistent, efficient, distributed publish/subscribe messaging system.

There are two low-level clients: The Producer and the Consumer:

##Producer example:

var Producer = require('Prozess').Producer;

var producer = new Producer('social', {host : 'localhost'});
producer.connect();
console.log("producing for ", producer.topic);
producer.on('error', function(err){
  console.log("some general error occurred: ", err);  
});
producer.on('brokerReconnectError', function(err){
  console.log("could not reconnect: ", err);  
  console.log("will retry on next send()");  
});

setInterval(function(){
  var message = { "thisisa" :  "test " + new Date()};
  producer.send(JSON.stringify(message), function(err){
    if (err){
      console.log("send error: ", err);
    } else {
      console.log("message sent");
    }
  });
}, 1000);

##Consumer example:

var Consumer = require('Prozess').Consumer;

var options = {host : 'localhost', topic : 'social', partition : 0, offset : 0};
var consumer = new Consumer(options);
consumer.connect(function(err){
  if (err) {  throw err; }
  console.log("connected!!");
  setInterval(function(){
    console.log("===================================================================");
    console.log(new Date());
    console.log("consuming: " + consumer.topic);
    consumer.consume(function(err, messages){
      console.log(err, messages);
    });
  }, 7000);
});

A Consumer can be constructed with the following options (default values as shown below):

var options = {
  topic: 'test',
  partition: 0,
  host: 'localhost',
  port: 9092,
  offset: null, // Number, String or BigNum
  maxMessageSize: Consumer.MAX_MESSAGE_SIZE,
  polling: Consumer.DEFAULT_POLLING_INTERVAL
};

##Documentation

var producer = new Producer(options)

type Message := String | Buffer

Producer := ({
  topic: String,
  partition?: Number,
  host?: String,
  port?: Number,
  connectionCache?: Boolean
}) => EventEmitter & {
  connect: () => void,
  send: (Array<Message> | Message, opts?: {
    partition?: Number,
    topic?: String
  }, cb: Callback<>) => void
}

To create a producer you must call Producer() with various options. The only required argument is the topic your producing to.

var Producer = require('prozess').Producer

var producer = new Producer({ topic: 'foos' })

options.topic

options.topic must be a String and is the topic in kafka that you will producer to

options.partition

options.partition is an optional Number and defaults to 0. You can specify this if you want to change which partition you publish to.

options.host and options.port

options.host determines the host location of the kafka node you are connecting to and options.port determines the port.

These default to "localhost" and 9092 which are the default kafka ports.

options.connectionCache

options.connectionCache is a Boolean you can set to opt in into connection caching. By default prozess will create one TCP connection per topic.

If you set options.connectionCache to true then prozess will use one TCP connection per host & port.

producer.connect(Callback)

You can call producer.connect(cb) to open your connection to kafka, the cb you pass in will be called once the connection is open.

You must call .connect() before calling .send()

producer.send(Message, opts?: Object, Callback)

To produce messages to kafka you should call .send() with either an array of String's or a single String.

You can pass .send() an optional options argument to customize the topic & partition for this single send() request.

You must also supply a Callback to handle any asynchronous errors.

##Installation:

 npm install prozess

##Checkout the code and run the tests:

 $ git clone https://github.com/cainus/Prozess.git
 $ cd Prozess ; make test-cov && open coverage.html

##Kafka Compatability matrix:

Kakfa 0.8.0 Release	Not Supported
Kafka 0.7.2 Release	Supported
Kafka 0.7.1 Release	Supported
Kafka 0.7.0 Release	Supported
kafka-0.6	Not Supported
kafka-0.05	Not Supported

Versions taken from http://incubator.apache.org/kafka/downloads.html