To implemente this challenge, I have decided to create the project from scratch. To do so, I have defined the following multi-modules project based on microservice architecture:
- weezam-registry: it is based on Spring Cloud Netflix Eureka server in order to implement the discovery pattern.
- weezam-gateway: it is based on Spring Cloud Gateway. This module will be responsible for expose the API.
- weezam-user: Microservice responsible for implementing endpoints which allow us to operate with Users & PaymentMethods.
- weezam-withdrawal: Microservice responsible for implementing the logic in order to support the withdrawal processing.
- weezam-notifications: Microservice responsible for processing the notifications from withdrawal processing.
Each of them have been implemented using a Hexagonal architecture with the following package structure:
- domain:
- model: It contains Model classes
- aggregate: It contains the business service classes
- repository: It contains Repository interfaces
- adapter:
- in
- amqp: It contains Message Broker Listener classes
- rest: It contains Rest API classes
- out
- amqp: It contains Message broker publisher classes
- persistence: It contains Repository implementation classes
- client: It contains Rest clients classes
- in
- Config: It contains spring config classes.
Also, I have prepared a docker-compose file in order to run all the modules together.
I am using RabbitMQ as JMS provider. Also, I am using Rabbitmq_delayed_message_exchange as plugin in order to scheduler withdrawals.
Apart from this, I have implemented the dead-letter queue + parking-lot queue approach in order to avoid losing some outgoing events about withdrawals
mvn clean package
docker-compose up --build
- Weezam Registry DASHBOAR
- RabbitMQ
- User:
guest - Password:
guest
- User:
I have only included some test cases in the weezam-user module. I apologise for that.