[KAFKA-10515] NPE: Foreign key join serde may not be initialized with default serde if application is distributed

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.6.0, 2.5.1
• Fix Version/s: 2.7.0, 2.6.1
• Component/s: streams
• Assignee: Thorsten Hake
• Reporter: Thorsten Hake
• Created: 23/Sep/20 08:50
• Updated: 12/Nov/20 10:35
• Resolved: 20/Oct/20 23:46

Description

The fix of KAFKA-9517 fixed the initialization of the foreign key joins serdes for KStream applications that do not run distributed over multiple instances.

However, if an application runs distributed over multiple instances, the foreign key join serdes may still not be initialized leading to the following NPE:

Encountered the following error during 
processing:java.lang.NullPointerException: null
	at 
org.apache.kafka.streams.kstream.internals.foreignkeyjoin.SubscriptionWrapperSerde$SubscriptionWrapperSerializer.serialize(SubscriptionWrapperSerde.java:85)
	at 
org.apache.kafka.streams.kstream.internals.foreignkeyjoin.SubscriptionWrapperSerde$SubscriptionWrapperSerializer.serialize(SubscriptionWrapperSerde.java:52)
	at 
org.apache.kafka.streams.state.internals.ValueAndTimestampSerializer.serialize(ValueAndTimestampSerializer.java:59)
	at 
org.apache.kafka.streams.state.internals.ValueAndTimestampSerializer.serialize(ValueAndTimestampSerializer.java:50)
	at 
org.apache.kafka.streams.state.internals.ValueAndTimestampSerializer.serialize(ValueAndTimestampSerializer.java:27)
	at 
org.apache.kafka.streams.state.StateSerdes.rawValue(StateSerdes.java:192)
	at 
org.apache.kafka.streams.state.internals.MeteredKeyValueStore.lambda$put$3(MeteredKeyValueStore.java:144)
	at 
org.apache.kafka.streams.processor.internals.metrics.StreamsMetricsImpl.maybeMeasureLatency(StreamsMetricsImpl.java:806)
	at 
org.apache.kafka.streams.state.internals.MeteredKeyValueStore.put(MeteredKeyValueStore.java:144)
	at 
org.apache.kafka.streams.processor.internals.ProcessorContextImpl$KeyValueStoreReadWriteDecorator.put(ProcessorContextImpl.java:487)
	at 
org.apache.kafka.streams.kstream.internals.foreignkeyjoin.SubscriptionStoreReceiveProcessorSupplier$1.process(SubscriptionStoreReceiveProcessorSupplier.java:102)
	at 
org.apache.kafka.streams.kstream.internals.foreignkeyjoin.SubscriptionStoreReceiveProcessorSupplier$1.process(SubscriptionStoreReceiveProcessorSupplier.java:55)
	at 
org.apache.kafka.streams.processor.internals.ProcessorNode.lambda$process$2(ProcessorNode.java:142)
	at 
org.apache.kafka.streams.processor.internals.metrics.StreamsMetricsImpl.maybeMeasureLatency(StreamsMetricsImpl.java:806)
	at 
org.apache.kafka.streams.processor.internals.ProcessorNode.process(ProcessorNode.java:142)
	at 
org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:201)
	at 
org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:180)
	at 
org.apache.kafka.streams.processor.internals.ProcessorContextImpl.forward(ProcessorContextImpl.java:133)
	at 
org.apache.kafka.streams.processor.internals.SourceNode.process(SourceNode.java:104)
	at 
org.apache.kafka.streams.processor.internals.StreamTask.lambda$process$3(StreamTask.java:383)
	at 
org.apache.kafka.streams.processor.internals.metrics.StreamsMetricsImpl.maybeMeasureLatency(StreamsMetricsImpl.java:806)
	at 
org.apache.kafka.streams.processor.internals.StreamTask.process(StreamTask.java:383)
	at 
org.apache.kafka.streams.processor.internals.AssignedStreamsTasks.process(AssignedStreamsTasks.java:475)
	at 
org.apache.kafka.streams.processor.internals.TaskManager.process(TaskManager.java:550)
	at 
org.apache.kafka.streams.processor.internals.StreamThread.runOnce(StreamThread.java:802)
	at 
org.apache.kafka.streams.processor.internals.StreamThread.runLoop(StreamThread.java:697)
	at 
org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:670)

This happens because the processors for foreign key joins will be distributed across multiple tasks. The serde will only be initialized with the default serde during the initialization of the task containing the sink node ("subscription-registration-sink"). So if the task containing the SubscriptionStoreReceiveProcessor ("subscription-receive") is not assigned to the same instance as the task containing the sink node, a NPE will be thrown because the Serde of the state store used within the SubscriptionStoreReceiveProcessor is not initialized.

Issue Links

• is related to: KAFKA-9517 KTable Joins Without Materialized Argument Yield Results That Further Joins NPE On
• links to: GitHub Pull Request #9338
• links to: GitHub Pull Request #9467