kafEEne #1: Websocket & Kafka

Say hello to kafEEne ! A blog series which will show Kafka and Java EE examples

This one is about Kafka + (Java EE) Websocket API. Code is on Github and you can refer to the README on how to get this up and running using Docker. Although the focus is on Websocket, here is a list of other Java EE specs which have been used – EJB, CDI & a bit of JSON-B (part of Java EE 8)

Websocket based dashboard

Consumer

Kafka Consumer is a @Stateless EJB
Initialized in the @PostConstruct callback
The consumer logic is executed within a while loop controlled by a (atomic) boolean flag which can potentially be mutated by the container. It sends the event data in form of a custom Payload object using CDI Events
Consumer shutdown is exposed within a @PreDestroy method where it invokes the wakeup method to interrupt the consumer while loop
A simple (one time) EJB timer job triggers the Consumer process – this is just for simplicity. No harm in exposing the consumer startup through, say, a REST endpoint

See Consumer for more details

Websocket endpoint

A server side Websocket endpoint which is actually does very little to be honest. Client can call it with the interested topic in the URL itself

The @OnOpen callback makes sure that the Websocket Session is added and tracked and the topic is added to the Session properties (Map)
@OnClose callback removes the Session from the memory

Broadcasting

It is enabled by a CDI observer (@Observe) for the Payload object
Simply loops over all the Sessions and broadcasts the key-value pair (received from the Kafka topic) if it registered for that topic to begin with
It uses the JSON-B API to convert the Payload POJO to a JSON format

See KafkaWebsocketEndpoint for more details

Connected clients

A @Singleton EJB maintains a list of connected clients
Its provides a synchronized wrapper over the collection of Websocket Session objects

See Peers for more details

Websocket Client

A simple Javascript based front end. See index.html for more details. I am not a front end guy – please pardon me!

Producer application

It’s a bare bones Java based producer application which pushes random data to couple of Kafka topics