How does Kafka Consumer Polling work when multiple partitions are assigned?

[jindalshivam09]

We want to have a Kafka consumer which should subscribe to multiple topics (each topic can have multiple partitions too). Now we also want to prioritize a particular topic such that say if there are two topics: topic_1 and topic_2, consumer should consume from topic_1 first and when all messages are consumed, then consumer from topic_2.

To implement such functionality, I started looking into how polling work in case of single topic but with multiple partitions are assigned to the consumer.
I have read various different answers over stack overflow and other sources, but still not sure with actual implementation. Could any please explain how is it implemented? How do we ensure fairness among partitions?

Also any ideas to implement priority among topics which a consumer is subscribed to multiple topics?
Couple of ideas I am thinking are:

1. Say if we have 100 partitions for topic_1, just create 2 partitions for topic_2, that way I can ensure that rate of consumption of topic_2 messages is very slow as compared to topic_1 (this is not exactly what we want though)
2. use pause() / resume() API i.e. pause topic_2 subscription, poll all the messages from topic_1, once done (or poll times out) resume topic_2 and consume few messages, pause again and continue with topic_1.