Why the library eats a bit of CPU despite lack of incoming messages?

Kafka Flow is validating if any of the timers need to be triggered from time to time during Kafka poll. If there are a lot of timers registered, the CPU usage could be noticable even if there are no incoming messages.

If CPU usage is too high, one might opt for configuring PartitionFlow to do such a triggering less often. See the PartitionFlow documentation in overview section for more details.