[KAFKA-10645] Forwarding a record from a punctuator sometimes it results in a NullPointerException

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.5.0
• Component/s: clients
• Assignee: Matthias J. Sax
• Reporter: Filippo Machi
• Created: 26/Oct/20 14:38
• Updated: 05/Nov/20 23:28

Description

Hello,
I am working on a java kafka stream application (v. 2.5.0) running on a kubernetes cluster.

It´s a springboot application running with java 8.

With the last upgrade to version 2.5.0 I started to see into the logs some NullPointerException that are happening when forwarding a record from a punctuator.
This is the stacktrace of the exception

Caused by: org.apache.kafka.streams.errors.StreamsException: task [2_2] Abort sending since an error caught with a previous record (timestamp 1603721062667) to topic reply-reminder-push-sender due to java.lang.NullPointerException\tat org.apache.kafka.streams.processor.internals.RecordCollectorImpl.send(RecordCollectorImpl.java:240)\tat org.apache.kafka.streams.processor.internals.RecordCollectorImpl.send(RecordCollectorImpl.java:111)\tat org.apache.kafka.streams.processor.internals.SinkNode.process(SinkNode.java:89)\tat org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:201)\tat org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:180)\tat org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:133)\t... 24 common frames omittedCaused by: java.lang.NullPointerException: null\tat org.apache.kafka.common.record.DefaultRecord.sizeOf(DefaultRecord.java:613)\tat org.apache.kafka.common.record.DefaultRecord.recordSizeUpperBound(DefaultRecord.java:633)\tat org.apache.kafka.common.record.DefaultRecordBatch.estimateBatchSizeUpperBound(DefaultRecordBatch.java:534)\tat org.apache.kafka.common.record.AbstractRecords.estimateSizeInBytesUpperBound(AbstractRecords.java:135)\tat org.apache.kafka.common.record.AbstractRecords.estimateSizeInBytesUpperBound(AbstractRecords.java:125)\tat org.apache.kafka.clients.producer.KafkaProducer.doSend(KafkaProducer.java:914)\tat org.apache.kafka.clients.producer.KafkaProducer.send(KafkaProducer.java:862)\tat org.apache.kafka.streams.processor.internals.RecordCollectorImpl.send(RecordCollectorImpl.java:181)\t... 29 common frames omitted

Checking the code, it looks like it happens calculating the size of the record. There is one header that is null but I don´t think I can control those headers right?
Thanks a lot

Issue Links

• links to: GitHub Pull Request #9518