tappoz / rust-personal-project

Rust backend architecture for REST APIs, shared libraries, DB and AMQP queues, CLIs, threads, TCP channels

Rust backend project

Overview

1. HTTP REST API to create/retrieve/search "work to do": pp_backend_api.
2. Postgres DB via Docker with a SQL schema: pp_storage.
3. pp_lib Rust library sharing the source code for the business logic.
4. CLI utility (cli_01) to interact with the HTTP API and the DB directly via pp_lib.
5. CLI utility (cli_02) to open a TCP socket and perform a manual HTTP call to pp_backend_api
6. AMQP RabbitMQ queue to publish/subscribe to produce/consume messages.
7. task_producer a schedule (CLI) to produce messages as "work demand" via AMQP.
8. A backend schedule to consume the AMQP queue - task_consumer that maps a "work demand" to a "work to do" in the DB, then stores "events" in the DB table to track the execution of the work to do, finally it updates the work row in the DB with the results of the calculations.

All the operations can be performed with a dedicated target in the Makefile.

The "work to do" is adding together numbers up to a upper bound threshold.

API payloads (HTTP REST):

{
  "id": 21,
  "work_code": "api-bjq8euwsEA",
  "add_up_to": 4,
  "done": false,
  "created_on": 1634115736,
  "updated_on": 1634115736
}

• Rust structure: Work.
• The work_code field has a prefix of api-*.
• The done field is false and is supposed to be updated when a hypothetical backend schedule picks up work to do from the database table works with a time-range filter (e.g. last 6 hours).
• Both the work_code suffix and the add_up_to field are randomly generated before adding the row to the database.

Queue messages (AMQP):

{
  "add_up_to": 2,
  "done": false
}

• Rust structure: WorkDemand.
• This is translated into a Rust structure Work by the task_consumer.
• The work_code for a message pulled from the queue has a prefix of consumer-*.
• The done field is updated to true once the calculations have been performed by the task_consumer.

These calculated rows can be searched for from the HTTP API to be retrieved.

Architecture