[KAFKA-8484] ProducerId reset can cause IllegalStateException

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.3.0
• Component/s: producer
• Assignee: Jason Gustafson
• Reporter: Jason Gustafson
• Created: 04/Jun/19 23:06
• Updated: 11/Jul/19 16:09
• Resolved: 12/Jun/19 19:57

Description

If the producerId is reset while inflight requests are pending, we can get the follow uncaught error.

[2019-06-03 08:20:45,320] ERROR [Producer clientId=producer-1] Uncaught error in request completion: (org.apache.kafka.clients.NetworkClient)                                                                                                               
java.lang.IllegalStateException: Sequence number for partition test_topic-13 is going to become negative : -965
        at org.apache.kafka.clients.producer.internals.TransactionManager.adjustSequencesDueToFailedBatch(TransactionManager.java:561)                                                                                                                      
        at org.apache.kafka.clients.producer.internals.Sender.failBatch(Sender.java:744)
        at org.apache.kafka.clients.producer.internals.Sender.failBatch(Sender.java:717)
        at org.apache.kafka.clients.producer.internals.Sender.completeBatch(Sender.java:667)
        at org.apache.kafka.clients.producer.internals.Sender.handleProduceResponse(Sender.java:574)
        at org.apache.kafka.clients.producer.internals.Sender.access$100(Sender.java:75)
        at org.apache.kafka.clients.producer.internals.Sender$1.onComplete(Sender.java:818)
        at org.apache.kafka.clients.ClientResponse.onComplete(ClientResponse.java:109)
        at org.apache.kafka.clients.NetworkClient.completeResponses(NetworkClient.java:561)
        at org.apache.kafka.clients.NetworkClient.poll(NetworkClient.java:553)
        at org.apache.kafka.clients.producer.internals.Sender.runOnce(Sender.java:335)
        at org.apache.kafka.clients.producer.internals.Sender.run(Sender.java:253)
        at java.lang.Thread.run(Thread.java:748)

The impact of this is that a failed batch will not be completed until the delivery timeout is exceeded. We are missing validation when we receive a produce response that the producerId and epoch still match.

Issue Links

• links to: Pull request