Hi, I’ve been trying to create hudi offline feature group for point-in-time query. But there is this error:
Seem like there is some “ts” columns being added. Am I doing this wrong?
Hi, I’ve been trying to create hudi offline feature group for point-in-time query. But there is this error:
Seem like there is some “ts” columns being added. Am I doing this wrong?
Hi,
it is not obvious, but you either need to provide a partition key or turn off partitioning by setting the following property in the Hudi arguments to an empty string:
import org.apache.hudi.DataSourceWriteOptions
val hudiArgs = Map(DataSourceWriteOptions.PARTITIONPATH_FIELD_OPT_KEY -> "")
(Hops.createFeaturegroup("hudi_test")
.setHudi(true)
.setDataframe(someDf)
.setHudiArgs(hudiArgs.asJava)
.setPrimaryKey(List("number_f").asJava).write())
Or provide a partition key:
(Hops.createFeaturegroup("hudi_test")
.setHudi(true)
.setDataframe(someDf)
.setPartitionBy(List("word_f").asJava)
.setPrimaryKey(List("number_f").asJava).write())
Append:
You can find more info here on the Hudi documentation in the Datasource Writer section.