[
https://issues.apache.org/jira/browse/KAFKA-4885?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15926571#comment-15926571
]
Guozhang Wang commented on KAFKA-4885:
--------------------------------------
[~damianguy] Well, as for the general solution in terms of error handling. I
think there is a difference between "streams app has an issue, e.g. divide by
zero, serde exception" and "broker cluster has an issue, and no data can be
produced / fetched". In the former case I agree that we should usually fail
fast; whereas for the latter case I'm not sure if we want Streams app to stop
and die whenever there is a (temporary?) broker unavailability. Imagine
cross-DC replication / MirrorMaker cases, we usually want such services to keep
alive and idle when brokers are unavailable rather than logging a fatal error
and shutdown itself.
As for the system test itself, I agree that setting a exception handler in
streams would be better than setting num.retries to infinity.
> processstreamwithcachedstatestore and other streams benchmarks fail
> occasionally
> ---------------------------------------------------------------------------------
>
> Key: KAFKA-4885
> URL: https://issues.apache.org/jira/browse/KAFKA-4885
> Project: Kafka
> Issue Type: Bug
> Components: streams
> Affects Versions: 0.10.2.0
> Reporter: Eno Thereska
> Fix For: 0.11.0.0
>
>
> test_id:
> kafkatest.benchmarks.streams.streams_simple_benchmark_test.StreamsSimpleBenchmarkTest.test_simple_benchmark.test=processstreamwithcachedstatestore.scale=2
> status: FAIL
> run time: 14 minutes 58.069 seconds
> Streams Test process on ubuntu@worker5 took too long to exit
> Traceback (most recent call last):
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/venv/local/lib/python2.7/site-packages/ducktape-0.6.0-py2.7.egg/ducktape/tests/runner_client.py",
> line 123, in run
> data = self.run_test()
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/venv/local/lib/python2.7/site-packages/ducktape-0.6.0-py2.7.egg/ducktape/tests/runner_client.py",
> line 176, in run_test
> return self.test_context.function(self.test)
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/venv/local/lib/python2.7/site-packages/ducktape-0.6.0-py2.7.egg/ducktape/mark/_mark.py",
> line 321, in wrapper
> return functools.partial(f, *args, **kwargs)(*w_args, **w_kwargs)
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/tests/kafkatest/benchmarks/streams/streams_simple_benchmark_test.py",
> line 86, in test_simple_benchmark
> self.driver[num].wait()
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/tests/kafkatest/services/streams.py",
> line 102, in wait
> self.wait_node(node, timeout_sec)
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/tests/kafkatest/services/streams.py",
> line 106, in wait_node
> wait_until(lambda: not node.account.alive(pid), timeout_sec=timeout_sec,
> err_msg="Streams Test process on " + str(node.account) + " took too long to
> exit")
> File
> "/var/lib/jenkins/workspace/system-test-kafka/kafka/venv/local/lib/python2.7/site-packages/ducktape-0.6.0-py2.7.egg/ducktape/utils/util.py",
> line 36, in wait_until
> raise TimeoutError(err_msg)
> TimeoutError: Streams Test process on ubuntu@worker5 took too long to exit
> The log contains several lines like:
> [2017-03-11 04:52:59,080] DEBUG Attempt to heartbeat failed for group
> simple-benchmark-streams-with-storetrue since it is rebalancing.
> (org.apache.kafka.clients.consumer.internals.AbstractCoordinator)
> [2017-03-11 04:53:01,987] DEBUG Sending Heartbeat request for group
> simple-benchmark-streams-with-storetrue to coordinator worker10:9092 (id:
> 2147483646 rack: null)
> (org.apache.kafka.clients.consumer.internals.AbstractCoordinator)
> [2017-03-11 04:53:02,088] DEBUG Attempt to heartbeat failed for group
> simple-benchmark-streams-with-storetrue since it is rebalancing.
> (org.apache.kafka.clients.consumer.internals.AbstractCoordinator)
> [2017-03-11 04:53:04,995] DEBUG Sending Heartbeat request for group
> simple-benchmark-streams-with-storetrue to coordinator worker10:9092 (id:
> 2147483646 rack: null)
> (org.apache.kafka.clients.consumer.internals.AbstractCoordinator)
> Other tests that fail the same way include:
> test_id:
> kafkatest.benchmarks.streams.streams_simple_benchmark_test.StreamsSimpleBenchmarkTest.test_simple_benchmark.test=count.scale=2
> test_id:
> kafkatest.benchmarks.streams.streams_simple_benchmark_test.StreamsSimpleBenchmarkTest.test_simple_benchmark.test=processstreamwithsink.scale=1
> test_id:
> kafkatest.tests.streams.streams_bounce_test.StreamsBounceTest.test_bounce
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
