Skip to content

study-notes.md

Latest commit

 

History

History
194 lines (100 loc) · 4.83 KB

study-notes.md

File metadata and controls

194 lines (100 loc) · 4.83 KB

STUDY NOTES: on RabbitMQ

tutorial --> https://www.youtube.com/watch?v=Cie5v59mrTg

RabbitMQ - open source, distributed messaging que
Written in ERLANG language, best suited for massively scalable soft real-time systems with high reliability
Supports many communication protocols
RabbitMQ Is a soft real-time system
It was originally trying to solve a problem called Spaghetti Mesh Architecture
The left figure is clients communicating to each other, the right is a queue service such as RabbitMQ https://i.ibb.co/xGjzsxN/Impact-N.png architecture



Rabbits own protocol = AMQP - Advanced message queue protocol


RabbitMQ Components: index

  • AMQP
  • RabbitMQ Server
  • Channels
  • Queue
  • Publisher
  • Consumer
  • Exchange
  • Protocols

https://i.ibb.co/GJwhpfY/RabbitMQ.png

Abstract view


RabbitMQ top level abstract view:

  1. The RabbitMQ server
  2. The Publisher
  3. The two way Stateful coms connection between the Publisher and RabbitMQ using AMQP
  4. The Consumer
  5. Another two way Stateful coms connection between the Consumer and RabbitMQ using AMQP
  6. Channels
  7. Exchanges


RabbitMQ Server

  • Rabbit uses a middle/Central layer that all clients communicate to

  • This solves the spaghetti mess

  • Could be multiple servers, could be distributed

  • Uses listens to port 5672 by default

  • It is using TCP so it has to listen

  • Pushes message from publishers to intended consumers


Publisher

  • “Hey I am the client and I want to publish a message to the consumers that are interested in this message”

  • Establishes a stateful TCP connection between itself(the publisher) and the RabbitMQ server

  • It is a 2 way/bi directional communication connection, therefor the underlying transportation protocol is raw TCP.

  • Not HTTP, but rather raw TCP

  • AMQP wraps around the TCP. This adds extra stuff like its own HEADERS

  • Publishers and Consumers are not aware of Queues. However they are aware of Exchanges

  • Publishers can freely be created and killed, unlike consumers who upon creation must stay alive


Consumer

  • Receives message from server
  • Must ‘acknowledge’ messages in order to clear them from the queue
  • Message acknowledgment is very complicated and gets extremely in-depth, because its so important
  • RabbitMQ guarantees 1 successful acknowledgment

Channels

  • Channels take 1 stateful TCP bidirectional connection between Publisher/Consumer and RabbitMQ server, and, splits it up.

  • For example, Let’s say a particular view/page on your frontend has 5 consumers, well, rather than each consumer creating a connection to Rabbit server, the client can create One connection, and share it among each consumers. Each consumers gets their “Channel” in the connection to the Rabbit server

  • Channels keep things separated. each consumers connection is separated from other consumers

  • The technical terminology for Channels is Multiplexing

  • HTTP/2 uses multiplexing


Exchanges

  • Exchanges take care of propagating messages into Queues
  • The default exchange works for most cases
  • Advanced uses. Kinda like Django middleware
  • Support algorithms/filters to ‘fan out’ to different queues and stuff

AMQP

  • Advanced Messaging Queue Protocol
  • Basically wraps raw TCP with extra stuff like HEADERS
  • Has ‘clients’,
  • In js, using the require(“amqplib”) , If anything returns a Bluebird, then that is a promise so must await it

How to RabbitMQ

  1. With docker installed

  2. run: docker run hello-world

  3. You should see Hello World! In terminal. This confirms docker is installed correctly

  4. Now run: docker run --name rabbitmq -p 5672:5672 rabbitmq

  5. docker run (a container) --name (immediately name your new container) rabbitmq -p (expose this port) port:port 5672:5672 rabbitmq(the actual rabbitmq container image that docker will pull from public container repository)

  6. When the image is downloaded and Spinned(span) up, that terminal will be ‘occupied’ as the running process of the container

  7. Create project folder rabbitmq

  8. Open folder in vscode

  9. Make publisher.js

  10. Run in terminal: npm init -y (creates package.json. -y tell npm “I know what I am doing just create package.json”)

Tips:

What we are going to build: https://i.ibb.co/9G6PFWg/Async-Job-Manager.png Async job manager

Asynchronous job execution manager/engine.

  • Publisher will publish jobs
  • The appropriate consumer for that job type will get the message and do the job