It is possibly because missing a SerDe propterty when you creating HIVE tables
for TPC-DS:
'serialization.null.format'=''
For example,
CREATE TABLE call_center
(
cc_call_center_sk BIGINT,
cc_call_center_id STRING,
cc_rec_start_date STRING,
cc_rec_end_date
This works fine
scala> sql("use oraclehadoop")
res1: org.apache.spark.sql.DataFrame = [result: string]
scala> sql("select count(1) from sales").show
+---+
|_c0|
+---+
|4991761|
+---+
You can do "select count(*) from tablename") as it is not dynamic sql. Does
it actually work?
Sin
