Kerberized kafka error

[Felixoid]

Changelog category (leave one):

• Build/Testing/Packaging Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

With the current approach, all ports are calculated at the beginning and could overlap or even be highjacked, see the report for port is already allocated. It's possibly the reason for #45368.

[yakov-olkhovskiy]

I'm afraid this approach will not guarantee reliable solution - port will be released between acquisition and real use, and it heavily depends on how we use these properties - for example we have code like this:

def kafka_produce(kafka_cluster, topic, messages, timestamp=None, retries=15):
    logging.debug(
        "kafka_produce server:{}:{} topic:{}".format(
            "localhost", kafka_cluster.kafka_port, topic
        )
    )
    producer = get_kafka_producer(
        kafka_cluster.kafka_port, producer_serializer, retries
    )
    for message in messages:
        producer.send(topic=topic, value=message, timestamp_ms=timestamp)
        producer.flush()

notice kafka_port is used in logging first...

[Felixoid]

I agree, it's better, but not 100% reliable.

It's much better at least by reducing the time between assigning *_port and using it. Another point, we don't collect every possible port at __init__ anymore, and reserve them only on demand