Using kafka.KafkaConsumer in multiple processes with the same group ID I'm noticing high CPU load when idle. I'm running the kafka-python 0.9 branch (commit 31c3d59) against Kafka 0.9.0.0. Group coordination is working as expected, with each message being processed by one consumer in the group, however (n-1)/n consumers in the group are running at 100% CPU. With n=1 CPU load is low as expected, however with n>1 I see n-1 processes running 100% CPU as if they were busy-spinning.

I'm seeing this with the following, stripped-down consumer worker function:

def consumer_worker(kafka_uri, topic, group_id, heartbeat, stats_array):
    consumer = kafka.KafkaConsumer(topic, group_id=group_id, bootstrap_servers=[kafka_uri])
    for message in consumer:
        print "consumer_worker iterating"

The print shows that the for isn't busy-spinning. I have not performed any further profiling at this point.