Hermes is a scoring engine written purely in Python. This project is a direct result of trying to easily score and track service uptime for Attack-Defense competitions. Hermes tries to be hip and modern by using Python 3, RabbitMQ + Celery, and probably some other things I am forgetting about.
Hermes uses RabbitMQ as a message broker for service checking. Celery workers from multiple locations can receive tasks from the queue which helps prevent bottlenecks. Docker has been the simplest way to run RabbitMQ, especially during development. This is the official RabbitMQ Docker repository and can be used out-of-the-box for this.
Workers are used by RabbitMQ to execute tasks. When a task has been sent to RabbitMQ, it is added into the queue. Then, the tasks are distributed to available workers. In order for Hermes to check services, there must be at least one worker connected.
Hermes utilizes Celery for workers and tasks
To start a single worker:
$ celery worker --app=app.mod_checkTo start, stop, and restart multiple workers:
# start 5 workers
$ celery multi start 5 --app=app.mod_check
# stop 5 workers
$ celery multi stop 5 --app=app.mod_check
# restart 5 workers
$ celery multi restart 5 --app=app.mod_checkTo use a different broker address or a different broker entirely, look at the
--broker option for celery.
There is also a script included called autoreload-worker.sh which will
automatically reload the celery worker if there have been code changes
(similar to running flask in debug mode). This will make the development
process much simpler as everything will automatically reload itself.
Generally, to run Hermes:
- Start RabbitMQ server
cd path/to/Hermescelery multi start 5 --app=app.mod_checkorbash autoreload-worker.shpython3 run.py