[BEAM-9652] BigQueryIO MultiPartitionsWriteTables fails with ClassCastException: java.lang.Object cannot be cast to org.apache.beam.sdk.values.KV

• Type: Bug
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Affects Version/s: 2.19.0
• Fix Version/s: 2.21.0
• Component/s: io-java-gcp
• Assignee: Luke Cwik
• Reporter: Luke Cwik
• Created: 31/Mar/20 22:15
• Updated: 16/May/20 14:19
• Resolved: 02/Apr/20 04:02

Description

It looks like the coder inference fails for BatchLoad.writeTempTables and selects an avro coder:

object_value: <
              type: "org.apache.beam.sdk.coders.AvroCoder"
              parameters: <
                name: "type"
                value: <
                  string_value: "java.lang.Object"
                >
              >
              parameters: <
                name: "schema"
                value: <
                  string_value: "{\"type\":\"record\",\"name\":\"Object\",\"namespace\":\"java.lang\",\"fields\":[]}"
                >
              >

Full exception:

exception: "java.lang.ClassCastException: java.lang.Object cannot be cast to org.apache.beam.sdk.values.KV at org.apache.beam.sdk.coders.KvCoder.registerByteSizeObserver(KvCoder.java:36) at org.apache.beam.sdk.coders.IterableLikeCoder.registerByteSizeObserver(IterableLikeCoder.java:191) at org.apache.beam.sdk.coders.IterableLikeCoder.registerByteSizeObserver(IterableLikeCoder.java:60) at org.apache.beam.sdk.util.WindowedValue$FullWindowedValueCoder.registerByteSizeObserver(WindowedValue.java:623) at org.apache.beam.sdk.util.WindowedValue$FullWindowedValueCoder.registerByteSizeObserver(WindowedValue.java:539) at org.apache.beam.runners.dataflow.worker.IntrinsicMapTaskExecutorFactory$ElementByteSizeObservableCoder.registerByteSizeObserver(IntrinsicMapTaskExecutorFactory.java:400)

Issue Links

• links to: GitHub Pull Request #11280