[2.9.1] With partitioned topics, the Python client returns UnknownError in some cases when it should return a specific error class

Describe the bug The Python client raises _pulsar.UnknownError: Pulsar error: UnknownError in cases when client logs indicate a specific exception code (for which there is already a specific error subclass) was recieved.

To Reproduce 0. Create a Client object with a Python logger that emits at INFO and above (e.g. logging.getLogger()).

1. Create a persistent topic with a backlog policy that is very “short”, e.g. 1mb.
2. With a Python producer, send messages to the topic continually until the backlog quota is hit.
3. Observe that the Python producer errors with an UnknownError rather than ProducerBlockedQuotaExceededException.
4. Observe that the client logs show the arrival of a ProducerBlockedQuotaExceededException event.
5. Attempt to re-create the producer on the quota-blocked topic.
6. Observe that the same issue occurs: logs indicate ProducerBlockedQuotaExceededException, but an UnknownError is raised.

Expected behavior Specific error classes (the exception classes distributed in the pulsar-client package) should be raised instead of UnknownError when the client can match an error event (visible in the logs) to an exception class: ProducerBlockedQuotaExceededException

I suspect the issue is with the boost-python mapping code; I don’t think class-type checks are matching correctly on C++ exception objects.

Example logs:

chariot.chariot.pulsar_client::75192::Dummy-8::WARNING 2022-04-07 18:06:35,902 [127.0.0.1:58048 -> 127.0.0.1:6650] Received error response from server: ProducerBlockedQuotaExceededException (Cannot create producer on topic with backlog quota exceeded) -- req_id: 0
chariot.chariot.pulsar_client::75192::Dummy-8::ERROR 2022-04-07 18:06:35,902 [persistent://chariot1/chariot_namespace_heartbeat/chariot_topic_heartbeat-partition-0, SynchronousChariotProducer-heartbeat-1-ZacBentleyMBP-75192] Failed to create producer: ProducerBlockedQuotaExceededException
chariot.chariot.pulsar_client::75192::Dummy-8::ERROR 2022-04-07 18:06:35,902 Unable to create Producer for partition - 0 Error - ProducerBlockedQuotaExceededException

Example stacktrace which accompanies those logs:

Traceback (most recent call last):
  File "/Users/zac.bentley/Desktop/Projects/Klaviyo/chariot/.venv/lib/python3.7/site-packages/pulsar/__init__.py", line 603, in create_producer
    p._producer = self._client.create_producer(topic, conf)
_pulsar.UnknownError: Pulsar error: UnknownError

[BewareMyPower]

I also reproduced this bug for a partitioned topic.

2022-04-11 16:29:20.714 INFO  [0x70000f276000] ProducerImpl:189 | [persistent://public/default/my-topic-partition-0, ] Created producer on broker [127.0.0.1:64964 -> 127.0.0.1:6650] 
2022-04-11 16:29:20.714 INFO  [0x70000f276000] ProducerImpl:662 | [persistent://public/default/my-topic-partition-0, standalone-0-7] Closing producer for topic persistent://public/default/my-topic-partition-0
2022-04-11 16:29:20.714 INFO  [0x70000f276000] ProducerImpl:662 | [persistent://public/default/my-topic-partition-1, ] Closing producer for topic persistent://public/default/my-topic-partition-1
2022-04-11 16:29:20.714 INFO  [0x70000f276000] ProducerImpl:662 | [persistent://public/default/my-topic-partition-2, ] Closing producer for topic persistent://public/default/my-topic-partition-2
2022-04-11 16:29:20.714 ERROR [0x70000f276000] PartitionedProducerImpl:151 | Unable to create Producer for partition - 0 Error - ProducerBlockedQuotaExceededException
2022-04-11 16:29:20.714 ERROR [0x11352f600] SampleProducer:35 | Error creating producer: UnknownError

I will fix this issue soon.

[zbentley]

When I create a truly non-partitioned topic, the behavior is correct (I get a ResultProducerBlockedQuotaExceededException on the client). So this bug only happens for partitioned topics. I’ll update the title accordingly.