Add metrics

[tvoinarovskyi]

• We have some research to do before. Here's the list of metrics http://kafka.apache.org/documentation.html#monitoring (correct if wrong). Select ones of interest, some don't have any meaning in asyncio context (like number of select calls).
• kafka-python has an implementation for metrics already. It would be best to reuse it to not support too much code. Some measurements are needed before thou.

[tvoinarovskyi]

Metrics are still not even documented in kafka-python. I think we should wait until it becomes a stable API there before adding those... For now I'll remove those from 0.2.0 milestone

[tvoinarovskyi]

It's not something too complicated, but requires some time. Basically we need to take the metrics aggregators present in kafka-python and call sensors for respectful events. Like self._sensors.records_fetch_lag.record(highwater - last_offset) after each fetch to record lag samples.

[Corfucinas]

Any progress? Just noticed the consumer does not have the method .metrics()

[jigar-bhati]

Any updates on this? Having producer/consumers metrics for aiokafka seems like a super useful thing to have for debugging client side issues or investigating client vs server problems.

[JaredCorduan]

Exposing metrics would be extremely valuable.

[marcosschroh]

We have the following metrics in our micro framework. The metrics are automatically generated, then you have to expose them in your prefer way. If you are using a web framework it is really easy, for example, otherwise you can use start_wsgi_server from prometheus_client

[JaredCorduan]

Thank you @marcosschroh , I will check this out now!