Test task: System health event notifier

test_treebo_task_4_health_event_notifier.md

System health event notifier

Problem setting

Suppose you have a remote system behind an API. The system is reactive: you send something to it, and it is expected to do something in result. There is an API endpoint (sink) that receives requests and a callback (source) that sends responses.

Sink is just a python function that eventually calls an http endpoint. Source could be implemented as webhook or a generator, but we'll focus on the latter case.

Let's say we have a notification channel that (n-channel) goes to humans, e.g. a telegram bot. Events are sent to n-channel using http api.

Task

Write a script that periodically sends events to sink and expects response from source. If no response received, send something to n-channel.

Specifics

This is not required, if you really want to implement for something else, go ahead.

For source and sink use kafka-python https://pypi.org/project/kafka-python/. Source = kafkaProducer, sink = kafkaConsumer.

For n-channel use a telegram bot. Principle:

def send_notification(message):
  r = requests.get('https://telegram.org/...')

Script inputs:

1. config for kafka-python: server and topic
2. config for telegram api: token and chat id where to send notifications.

Criterions

1. Readability
2. Extensibility and ability to modify
3. Attention to edge cases. Everything that can possibly happen will happen.
4. Ability to decide things that are not clear on yourself. Not all tasks are clear and sometimes you have to a) guess what people mean b) ask.