fastify-redis-channels
v1.3.1
Published
A Fastify plugin for channels implementation based on Redis streams.
Downloads
1,373
Maintainers
eusunovReadme
fastify-redis-channels
A Fastify plugin for fast, reliable, and scalable message broker implementation based on Redis streams.
Suitable for IoT applications with massive network traffic, pub/sub-use cases, or any implementation with multiple producers/consumers.
Implements a possibility to scale the message processing across different consumers, without single consumers having to process all the messages. A group of consumers, working as a team, can cooperate and consume a separate portion of those messages from the same channel.
Can be used with a single Redis instance and later updated easily to a cluster configuration without need of any application change.
The implementation uses native Promises.
Do you want your project to grow? Then start right from the begging.
Under the hood @hearit-io/redis-channels is used, the options you pass to a register will be passed to the RedisChannels instance.
Table of Contents
Install
$ npm install fastify-redis-channels --saveUsage
Add it to your project with register and that's it!
You can access the RedisChannels instance via fastify.channels. A RedisChannelsError object is accessible via fastify.RedisChannelsError.
If you want to define a separate RedisChannels instace for a given namespace you can add a namespace property in the options. You can access the namespaced instance via fastify.channels[namespace]
All channels are automatically closed when a fastify instance is closed.
Learn by doing
Chat server
We will create a basic chat server based on websockets in this example.
Step 1 - Install all required packages
Create an empty folder for your application and initialise it:
mkdir chat
cd chat
npm initInstall all required packages:
npm install --save fastify fastify-websocket fastify-redis-channelsStep 2 - Create a chat room page
This step implements a chat room page where a user can send/receive messages to/from other users.
It automatically creates a chat room for everything in the root path of the URL.
For example, visiting http://localhost/room1 and http://localhost/room2 will create chat rooms room1 and room2.
Create a room.js file with a following content:
'use strict'
function fastifyPluginRoom (fastify, opts, done) {
// Route to the room
fastify.get('/:room', (request, reply) => {
reply.type('text/html').send(view(request.params.room))
})
done()
}
// Builds a page view with a text area, input field and a submit button.
function view (room) {
const page = `
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8"/>
<title>Chat room '${room}'</title>
</head>
<body>
<textarea id="log" cols="100" rows="20" readonly></textarea><br>
<input id="input" type="text" size="100"><br>
<input id="submit" type="button" value="Send"> to room '${room}'
<script>
const ws = new WebSocket(
'ws://' + window.location.host + '/ws/' + '${room}'
)
ws.onmessage = function(e) {
const data = JSON.parse(e.data)
document.querySelector('#log').value += (data.message + '\\n')
}
ws.onclose = function(e) {
console.error('socket closed')
}
document.querySelector('#input').focus()
document.querySelector('#input').onkeyup = function(e) {
if (e.keyCode === 13) {
document.querySelector('#submit').click()
}
}
document.querySelector('#submit').onclick = function(e) {
const inputElem = document.querySelector('#input')
ws.send(JSON.stringify({ 'message': inputElem.value }))
inputElem.value = ''
}
</script>
</body>
</html>
`
return page
}
module.exports = fastifyPluginRoom
Step 3 - Create a Fastify server
In this step we implement a simple Fastify server listening on port 3000.
Create a file server.js as shown bellow:
'use strict'
const fastify = require('fastify')()
fastify.register(require('./room'))
fastify.ready(error => {
if (error) console.log(error)
})
fastify.listen({ port: 3000 }, (error, address) => {
if (error) console.log(error)
console.log('Listen on : ', address)
})Run the server with the command:
npm startYou will get on the console the following output:
Listen on : http://[::]:3000You should see your chat room on http://localhost:3000/room
Step 4 - Create a consumer using channels
In this step we will create a consumer which broadcasts all messages received via websockets to all clients in the corresponding chat rooms.
Create a file consumer.js as shown below:
'use strict'
function fastifyConsumerPlugin(fastify, opts, done) {
try {
fastify.get('/ws/:room', { websocket: true }, handler)
done()
} catch(error) {
done(error)
}
}
// A websocket handle function (called once after a handshake)
async function handler(connection, req, params) {
const fastify = this
try {
// Creates a tunnel object to access a channel associated with the room.
const tunnel = await fastify.channels.use(params.room)
// Subscribes for messages.
await fastify.channels.subscribe(tunnel)
// Starts a consumer.
consume(fastify, connection, tunnel)
.then((resolve) => {
console.log('Consumer finished')
})
.catch((reject) => {
connection.socket.close()
return
})
// Produces received from a websocket messages to the corresponding tunnel.
connection.socket.on('message', async (message) => {
try {
connection.resume()
await fastify.channels.produce(tunnel, message)
} catch (error) {
connection.socket.close()
return
}
})
// Unsubscribe on websocket close
connection.socket.on('close', async () => {
await fastify.channels.unsubscribe(tunnel)
})
}
catch(error) {
connection.socket.close()
}
}
// A consumer implementation
// Consumes messages from the tunnel and broadcast them to the websocket.
async function consume(fastify, connection, tunnel) {
for await (const messages of fastify.channels.consume(tunnel)) {
for (const i in messages) {
connection.socket.send(messages[i].data)
}
}
}
module.exports = fastifyConsumerPluginRegister all plugins in the fastify server. The file server.js should look like this:
'use strict'
const fastify = require('fastify')()
fastify.register(require('fastify-websocket'))
fastify.after(error => {
if (error) console.log(error)
})
fastify.register(require('fastify-redis-channels'), {
channels: {
application: 'example',
},
redis: {
host: 'localhost',
port: 6379
}
})
fastify.after(error => {
if (error) console.log(error)
})
fastify.register(require('./consumer'))
fastify.after(error => {
if (error) console.log(error)
})
fastify.register(require('./room'))
fastify.ready(error => {
if (error) console.log(error)
})
fastify.listen({ port: 3000 }, (error, address) => {
if (error) console.log(error)
console.log('Listen on : ', address)
})Step 5 - Test your chat application
Before you test your chat, room application make sure you have up and running Redis server on the default host localhost and port 6379. For more info about the installation see on Redis download page.
Run the server with the command:
npm startYou will get on the console the following output:
Listen on : http://[::]:3000Open in two browser window a link to our example chat room http://localhost:3000/room.
Have a fun with your chat! :)
The complete example is available here fastify-redis-channels-chat-example.
Statistic worker
In this example, our web server for each request will send through a channel statistic a data (user agent and the IP address) to a separate worker process for a further processing. This will offload a server from a resource expensive operations.
Step 1 - Install all required packages
Create an empty folder for your application and initialise it:
mkdir worker
cd worker
npm initInstall all required packages:
npm install --save fastify fastify-redis-channels @hearit-io/redis-channelsStep 2 - Create a Fastify server
In this step we implement a simple Fastify server listening on port 3000.
For each requiest on a route / the server will produce a message in the channel statistic.
Create a file server.js as shown bellow:
'use strict'
const fastify = require('fastify')()
fastify.register(require('fastify-redis-channels'), {
channels: {
application: 'worker',
},
redis: {
host: 'localhost',
port: 6379
}
})
fastify.ready(error => {
if (error) console.log(error)
})
fastify.get('/', async (request, reply) => {
// Produces a statistic message for each request
const tunnel = await fastify.channels.use('statistic')
const message = {
agent: request.headers['user-agent'],
ip: request.ip
}
fastify.channels.produce(tunnel, JSON.stringify(message))
reply.type('text/html').send('Hello World')
})
fastify.listen({ port: 3000 }, (error, address) => {
if (error) console.log(error)
console.log('Listen on : ', address)
})
Step 3 - Create a worker process
In this step we will implement a consumer worker process with two consumers (working in a team).
Create a file worker.js as shown bellow:
'use strict'
const {RedisChannels} = require('@hearit-io/redis-channels')
// The channels instance
const channles = new RedisChannels({
channels: {
application: 'worker',
},
redis: {
host: 'localhost',
port: 6379
}
})
// Handle Control-D, Control-C
async function handle(signal) {
const channels = this
await channels.cleanup()
}
process.on('SIGINT', handle.bind(channles))
process.on('SIGTERM', handle.bind(channles))
process.on('SIGHUP', handle.bind(channles))
// A consumer function
async function consume(tunnel, channels) {
for await (const messages of channels.consume(tunnel)) {
for (const i in messages) {
// Process a message
const data = JSON.parse(messages[i].data)
console.log(data)
}
}
}
// The main loop
async function main () {
try {
// Creates tunnels to 'statistic'
const tunnelConsumerOne = await channles.use('statistic')
const tunnelConsumerTwo = await channles.use('statistic')
// Subscribe consumers in team
await channles.subscribe(tunnelConsumerOne, 'team')
await channles.subscribe(tunnelConsumerTwo, 'team')
// Start all consumers
consume(tunnelConsumerOne, channles).catch(() => { })
consume(tunnelConsumerTwo, channles).catch(() => { })
}
catch (error) {
console.log(error)
}
}
main()
Step 4 - Test a statistic worker
In order to test the worker open two console windows and start a Fastify server and a worker process.
Start the server:
npm startStart the worker process:
node worker.jsA request to http://localhost:3000/ wil result in a worker console output like this:
{
agent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36 Edg/86.0.622.69',
ip: '::ffff:78.83.64.18'
}
{
agent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.198 Safari/537.36',
ip: '::ffff:78.83.64.18'
}
The complete example is available here fastify-redis-channels-worker-example.
Service Send Events example
This example presents a real use case of SSE similar to our HEARIT.IO implementation. See here the complete sources.
The live SSE frontend implementation from our HEARIT.IO project is available here.
It demonstrates a reliable way of delivering service send events by using Redis stream ids and detecting gaps.
Project status
hearit.io
| Smart home automatization designed for visually impaired people. ------------ | -------------
@heart-io/redis-channels is used productive in our progressive web app. You can try it with a user [email protected] and password: 11223344
The package will be updated and maintained regularly.
The main goal of hearit.io is to make accessible the world of IoT to everyone. We created a speaking home suitable for all.
We will be grateful to you if you make awareness to other people of our project.
To finance our idea, we would be glad to support and work on your projects. Contact Emil [email protected] for details and our availability.
Other open-source packages, part of our project, will be available soon. We use Fastify as an application framework.
Todo
- [x] Add plugin unit tests.
- [x] Add a possibility to have multiple channels instances (per
namespace). - [ ] TypeScript support.
Authors and acknowledgment
Emil Usunov
License
MIT