Reply
Highlighted
Expert Contributor
Posts: 139
Registered: ‎07-21-2014

Kite dataset partition-by causing AvroTypeException on read

I created a dataset with a partition strategy with type 'year', 'month', 'day' based on timestamp and was able to write to the dataset using Flume's Kite dataset sink.

 

But on read using HIve, I notice this exception for "select * from <tbl> limit 10;" query:

 

java.io.IOException: org.apache.avro.AvroTypeException: Found logs_table,
expecting logs_table, missing required field typeid

 

typeid is not a required field and has a default null defined. If I create the dataset without the partitioning strategy, I'm able to run the query and get the expected results.

 

Any input on what might be going wrong? Thanks!