David Wayne Birdsall created TRAFODION-3148:
-----------------------------------------------
Summary: Creating a persistent sample table on a Hive table fails
with some columns
Key: TRAFODION-3148
URL: https://issues.apache.org/jira/browse/TRAFODION-3148
Project: Apache Trafodion
Issue Type: Bug
Components: sql-cmp
Affects Versions: 2.3
Reporter: David Wayne Birdsall
Assignee: David Wayne Birdsall
The following reproduces the problem.
First, create and populate a Hive table that uses the column name
"_DIVISION_1_". (In Hive, one uses backticks to define delimited column names
rather than double quotes as in standard SQL.):
{{hive> create table dave2( a int, `_DIVISION_1_` int)}}
{{ > stored as textfile}}
{{ > ;}}
{{OK}}
{{Time taken: 0.122 seconds}}
{{hive> insert into dave2 values (1,2),(3,4),(5,6),(7,8);}}
Then, in Trafodion sqlci, attempt to create a random sample. This will fail:
{{>>update statistics for table hive.hive.dave2 create sample random 90
percent;}}
{{*** ERROR[9200] UPDATE STATISTICS for table HIVE.HIVE.DAVE2 encountered an
error (4023) from statement Process_Query.}}
{{*** ERROR[4023] The degree of each row value constructor (1) must equal the
degree of the target table column list (2).}}{{*** ERROR[8822] The statement
was not prepared.}}{{--- SQL operation failed with errors.}}
{{>>}}
The probable cause of the problem is that "_DIVISION_1_" is assumed in
Trafodion to be a derived column, but this assumption is incorrect for Hive
tables.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
