[
https://issues.apache.org/jira/browse/KAFKA-15259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17748555#comment-17748555
]
Tomonari Yamashita commented on KAFKA-15259:
--------------------------------------------
Hi [~mjsax],
I think the stacktrace can be ignored. This problem cause is that sending a too
large message to "output-topic", then RecordTooLargeException occurs in
producer.send(), and the transaction would be rollbacked/aborted (3) despite
ProductionExceptionHandlerResponse.CONTINUE if using execute_once. The
stacktrace is just secondary disaster with an error caused by a subsequent
commit on the transaction that has already been rollbacked/aborted.
As far as we have tested, even using the same source code (attached file:
Reproducer.java), changing the version of Kafka Streams will lead to a
different behavior whether "continue" is available or not (i.e., the
transaction aborts), as follows:
- Kafka 2.6.3 : OK. "continue" worked as I assume.
- Kafka 2.8.2 : OK. "continue" worked as I assume.
- Kafka 3.0.0 : OK. "continue" worked as I assume.
- Kafka 3.1.2 : OK. "continue" worked as I assume.
- Kafka 3.2.0 : NG. "continue" didn't work because the transaction was
rollbacked/aborted.
- Kafka 3.2.3 : NG. "continue" didn't work because the transaction was
rollbacked/aborted.
- Kafka 3.5.1 : NG. "continue" didn't work because the transaction was
rollbacked/aborted.
>From what I have seen in the debugger, it is probably transaction aborting at
>those codes(3).
Probably due to this change (4)(5) in Kafka 3.2.0, if any of the individual
producer.send() fails, the transaction is aborted forcibly.
Since this change (4) (5) is in Kafka producer, I realize that it would be
difficult to achieve the behavior prior to 3.1.2, which allows "continue" to be
used. However, if "continue" is not available, there is a concern that Kafka
Streams applications will continue to get stuck because there is no way to
ignore messages that raise RecordTooLargeException by sending them to the dead
letter queue.
(3)
https://github.com/apache/kafka/blob/3.5.1/clients/src/main/java/org/apache/kafka/clients/producer/KafkaProducer.java#L1081-L1083
{code:java}
if (transactionManager != null) {
transactionManager.maybeTransitionToErrorState(e);
}
{code}
(4) Silent data loss in Kafka producer
https://issues.apache.org/jira/browse/KAFKA-9279
(5) KAFKA-9279: Fail producer transactions for asynchronously-reported,
synchronously-encountered ApiExceptions #11508
https://github.com/apache/kafka/pull/11508
> Kafka Streams does not continue processing due to rollback despite
> ProductionExceptionHandlerResponse.CONTINUE if using execute_once
> ------------------------------------------------------------------------------------------------------------------------------------
>
> Key: KAFKA-15259
> URL: https://issues.apache.org/jira/browse/KAFKA-15259
> Project: Kafka
> Issue Type: Bug
> Components: streams
> Affects Versions: 3.5.1
> Reporter: Tomonari Yamashita
> Priority: Major
> Attachments: Reproducer.java, app_at_least_once.log,
> app_exactly_once.log
>
>
> [Problem]
> - Kafka Streams does not continue processing due to rollback despite
> ProductionExceptionHandlerResponse.CONTINUE if using execute_once.
> -- "CONTINUE will signal that Streams should ignore the issue and continue
> processing"(1), so Kafka Streams should continue processing even if using
> execute_once when ProductionExceptionHandlerResponse.CONTINUE used.
> -- However, if using execute_once, Kafka Streams does not continue
> processing due to rollback despite
> ProductionExceptionHandlerResponse.CONTINUE. And the client will be shut down
> as the default behavior(StreamThreadExceptionResponse.SHUTDOWN_CLIENT)
> [Environment]
> - Kafka Streams 3.5.1
> [Reproduction procedure]
> # Create "input-topic" topic and "output-topic"
> # Put several messages on "input-topic"
> # Execute a simple Kafka streams program that transfers too large messages
> from "input-topic" to "output-topic" with execute_once and returns
> ProductionExceptionHandlerResponse.CONTINUE when an exception occurs in the
> producer. Please refer to the reproducer program (attached file:
> Reproducer.java).
> # ==> However, Kafka Streams does not continue processing due to rollback
> despite ProductionExceptionHandlerResponse.CONTINUE. And the stream thread
> shutdown as the default
> behavior(StreamThreadExceptionResponse.SHUTDOWN_CLIENT) (2). Please refer to
> the debug log (attached file: app_exactly_once.log).
> ## My excepted behavior is that Kafka Streams should continue processing
> even if using execute_once. when ProductionExceptionHandlerResponse.CONTINUE
> used.
> [As far as my investigation]
> - FYI, if using at_least_once instead of execute_once, Kafka Streams
> continue processing without rollback when
> ProductionExceptionHandlerResponse.CONTINUE is used. Please refer to the
> debug log (attached file: app_at_least_once.log).
> - "continue" worked in Kafka Streams 3.1.2, but no longer works since Kafka
> Streams 3.2.0, as rollback occurs.
> (1) CONFIGURING A STREAMS APPLICATION > default.production.exception.handler
> -
> [https://kafka.apache.org/35/documentation/streams/developer-guide/config-streams.html#default-production-exception-handler]
> (2) Transaction abort and shutdown occur
> {code:java}
> 2023-07-26 21:27:19 DEBUG KafkaProducer:1073 - [Producer
> clientId=java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1-0_0-producer,
> transactionalId=java-kafka-streams-0_0] Exception occurred during message
> send:
> org.apache.kafka.common.errors.RecordTooLargeException: The message is
> 11000088 bytes when serialized which is larger than 1048576, which is the
> value of the max.request.size configuration.
> 2023-07-26 21:27:19 ERROR RecordCollectorImpl:322 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> stream-task [0_0] Error encountered sending record to topic output-topic for
> task 0_0 due to:
> org.apache.kafka.common.errors.RecordTooLargeException: The message is
> 11000088 bytes when serialized which is larger than 1048576, which is the
> value of the max.request.size configuration.
> Exception handler choose to CONTINUE processing in spite of this error but
> written offsets would not be recorded.
> org.apache.kafka.common.errors.RecordTooLargeException: The message is
> 11000088 bytes when serialized which is larger than 1048576, which is the
> value of the max.request.size configuration.
> 2023-07-26 21:27:19 INFO TransactionManager:393 - [Producer
> clientId=java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1-0_0-producer,
> transactionalId=java-kafka-streams-0_0] Transiting to abortable error state
> due to org.apache.kafka.common.errors.RecordTooLargeException: The message is
> 11000088 bytes when serialized which is larger than 1048576, which is the
> value of the max.request.size configuration.
> 2023-07-26 21:27:19 DEBUG TransactionManager:986 - [Producer
> clientId=java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1-0_0-producer,
> transactionalId=java-kafka-streams-0_0] Transition from state IN_TRANSACTION
> to error state ABORTABLE_ERROR
> org.apache.kafka.common.errors.RecordTooLargeException: The message is
> 11000088 bytes when serialized which is larger than 1048576, which is the
> value of the max.request.size configuration.
> 2023-07-26 21:27:19 DEBUG StreamThread:825 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Processed 1 records with 1 iterations; invoking punctuators if necessary
> 2023-07-26 21:27:19 DEBUG StreamThread:837 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1] 0
> punctuators ran.
> 2023-07-26 21:27:19 DEBUG StreamThread:1117 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Committing all active tasks [0_0] and standby tasks [] since 273ms has
> elapsed (commit interval is 100ms)
> 2023-07-26 21:27:19 DEBUG RecordCollectorImpl:345 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> stream-task [0_0] Flushing record collector
> 2023-07-26 21:27:19 DEBUG StreamTask:419 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1] task
> [0_0] Prepared RUNNING task for committing
> 2023-07-26 21:27:19 DEBUG TaskExecutor:176 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Committing task offsets {0_0={input-topic-0=OffsetAndMetadata{offset=629,
> leaderEpoch=null, metadata='AgAAAYmSGuLv'}}}
> 2023-07-26 21:27:19 ERROR KafkaStreams:537 - stream-client
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d] Encountered the
> following exception during processing and the registered exception handler
> opted to SHUTDOWN_CLIENT. The streams client is going to shut down now.
> org.apache.kafka.streams.errors.StreamsException: Error encountered trying to
> commit a transaction [stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> stream-task [0_0]]
> at
> org.apache.kafka.streams.processor.internals.StreamsProducer.commitTransaction(StreamsProducer.java:313)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.TaskExecutor.commitOffsetsOrTransaction(TaskExecutor.java:186)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.TaskExecutor.commitTasksAndMaybeUpdateCommittableOffsets(TaskExecutor.java:154)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.TaskManager.commitTasksAndMaybeUpdateCommittableOffsets(TaskManager.java:1678)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.TaskManager.commit(TaskManager.java:1646)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.StreamThread.maybeCommit(StreamThread.java:1121)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.StreamThread.runOnce(StreamThread.java:840)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.StreamThread.runLoop(StreamThread.java:617)
> ~[kafka-streams-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:579)
> [kafka-streams-3.5.1.jar:?]
> Caused by: org.apache.kafka.common.KafkaException: Cannot execute
> transactional method because we are in an error state
> at
> org.apache.kafka.clients.producer.internals.TransactionManager.maybeFailWithError(TransactionManager.java:1010)
> ~[kafka-clients-3.5.1.jar:?]
> at
> org.apache.kafka.clients.producer.internals.TransactionManager.sendOffsetsToTransaction(TransactionManager.java:306)
> ~[kafka-clients-3.5.1.jar:?]
> at
> org.apache.kafka.clients.producer.KafkaProducer.sendOffsetsToTransaction(KafkaProducer.java:757)
> ~[kafka-clients-3.5.1.jar:?]
> at
> org.apache.kafka.streams.processor.internals.StreamsProducer.commitTransaction(StreamsProducer.java:300)
> ~[kafka-streams-3.5.1.jar:?]
> ... 8 more
> Caused by: org.apache.kafka.common.errors.RecordTooLargeException: The
> message is 11000088 bytes when serialized which is larger than 1048576, which
> is the value of the max.request.size configuration.
> 2023-07-26 21:27:19 INFO KafkaStreams:340 - stream-client
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d] State transition
> from RUNNING to PENDING_ERROR
> 2023-07-26 21:27:19 INFO StreamThread:239 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> State transition from RUNNING to PENDING_SHUTDOWN
> 2023-07-26 21:27:19 INFO StreamThread:1182 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Shutting down unclean
> 2023-07-26 21:27:19 INFO StreamThread:1168 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Informed to shut down
> 2023-07-26 21:27:19 DEBUG StreamThread:224 - stream-thread
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1]
> Ignoring request to transit from PENDING_SHUTDOWN to PENDING_SHUTDOWN: only
> DEAD state is a valid next state
> 2023-07-26 21:27:19 INFO KafkaStreams:1367 - stream-client
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d] Shutting down 1
> stream threads
> 2023-07-26 21:27:19 DEBUG KafkaStreams:1374 - stream-client
> [java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d] Shutdown
> java-kafka-streams-e3187cf9-5337-4155-a7cd-fd4e426b889d-StreamThread-1
> complete
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
