This project is my solution for the Kisi Backend Code Challenge .
The project contains a simple background job system that is compatible with the ActiveJob interface of Rails and
allows Rails developers to easily enqueue jobs to a Google Pub/Sub backend. Additionally, a rake task runs a process
that executes the enqueued jobs. Both the enqueue and the process happens on a specific queue.
A queue is represented by a Topic-Subscription pair with the same name in the Pub/Sub system. If none exists,
the engine will automatically create both the Topic and Subscription objects.
Pass the PUBSUB_JOB_QUEUE_PROJECT_ID variable to set the id of the Pub/Sub project holding Topics
and Subscriptions for the job queues. Make sure that both the web and worker docker services get the same value.
If not defined, job_queue will be set as a project id.
Pass the QUEUE_NAME variable to set the name of the job queue for a new worker. If not defined,
the queue name will be set to "default".
To start all services, make sure you have Docker installed and run:
$ docker compose up
To test the background job system, open a rails console and enqueue MockJob jobs:
$ docker compose exec web rails console
> MockJob.perform_later(sleep_duration: 3, raise_error: true)
The example above will start a job that sleeps 3 seconds, then raises a JobError.
There is retry_on configured on this error, so the ActiveJob system will re-enqueue the job 2 more times,
five minutes apart. You can change this configuration in the ApplicationJob class.
You can use a job spawner service to quickly enqueue a number of jobs:
$ docker compose exec web rails console
> Pubsub::Job::Spawner.call(job_count: 10)
The example above will enqueue 10 MockJob jobs where both the sleep_duration and the raise_error
values are randomized.
You can remove all pending jobs from a queue with the worker:clear task:
$ docker compose exec web rake worker:clear
To execute the test suite, run:
$ docker compose run --rm web rails test
If you run docker with a VM (e.g. Docker Desktop for Mac), it is recommended to allocate at least 2GB Memory.