K. Chen
K. Chen

Reputation: 46

Write Flink DataStream to Iceberg Table:NoSuchMethodError: org.apache.parquet.schema.Types$PrimitiveBuilder.as

I try to write a flink datastream to a iceberg table, as below:

val kafkaStream = new KafkaDataSource(parameter, new PacketSchema).getStream(env)
val dataStream = kafkaStream.flatMap(new NullPacketFilter).map(FilteredPacket.from(_).toRow).javaStream

FlinkSink.forRow(dataStream, FilteredPacket.schema)
  .tableLoader(tableLoader)
  .build

then I get error:

2021-02-18 18:12:12,086 WARN  org.apache.hadoop.util.NativeCodeLoader [] - Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
2021-02-18 18:12:12,424 INFO  org.apache.iceberg.BaseMetastoreCatalog [] - Table loaded by catalog: iceberg.flink_test.filtered_packets
2021-02-18 18:12:12,477 WARN  org.apache.flink.runtime.taskmanager.Task [] - Source: tianyi -> Flat Map -> Map -> Map -> IcebergStreamWriter (1/1) (9612408d42df7e69b829367434bbc43d) switched from RUNNING to FAILED.
java.lang.NoSuchMethodError: org.apache.parquet.schema.Types$PrimitiveBuilder.as(Lorg/apache/parquet/schema/LogicalTypeAnnotation;)Lorg/apache/parquet/schema/Types$Builder;
   at org.apache.iceberg.parquet.TypeToMessageType.primitive(TypeToMessageType.java:145) ~[tianyi112-1.0-SNAPSHOT.jar:?]
   at org.apache.iceberg.parquet.TypeToMessageType.field(TypeToMessageType.java:88) ~[tianyi112-1.0-SNAPSHOT.jar:?]
   at org.apache.iceberg.parquet.TypeToMessageType.convert(TypeToMessageType.java:65) ~[tianyi112-1.0-SNAPSHOT.jar:?]
   at org.apache.iceberg.parquet.ParquetSchemaUtil.convert(ParquetSchemaUtil.java:43) ~[tianyi112-1.0-SNAPSHOT.jar:?]
   ...

I have checked the content of jar, it includes the required Types$PrimitiveBuilder class:

2651 Fri Feb 19 08:32:10 CST 2021 org/apache/parquet/schema/Types$PrimitiveBuilder.class
3101 Fri Feb 19 08:32:12 CST 2021 org/apache/flink/hive/shaded/parquet/schema/Types$PrimitiveBuilder.clas

when I check the source code, found error in idea:

Library Source does not match the bytecode for class TypeToMessageType

but all other class is ok.

I tried to delete iceberg-parquet.jar and parquet-column.jar in my maven repository and reimport project, and tried to disable Idea's Lombok plugin - but has no effect.

version: CDH 6.3.2 Flink 1.11.2 Iceberg 0.11.0

Upvotes: 0

Views: 2392

Answers (1)

K. Chen
K. Chen

Reputation: 46

I've resolved the problem. The reason is jar conflicts, specifically, the parquet-hadoop.jar(hive-exec 2.3.4) and iceberg-parquet.jar(0.11.0)

Upvotes: 1

Related Questions