Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Invalid
-
None
-
None
Description
git.commit.id.abbrev=01bf849
I have a hbase table that contains empty values for some columns/rows. This was due to importing data from a file into hbase. For example a get of a row containing empty age value in hbase:
hbase(main):003:0> get 'voterspaces', '17'
COLUMN CELL
fourcf:create_date timestamp=1401380832939, value=2014-06-04 06:49:06
onecf:name timestamp=1401380832939, value=alice garcia
threecf:contributions timestamp=1401380832939, value=468.51
threecf:voterzone timestamp=1401380832939, value=18555
twocf:age timestamp=1401380832939, value=
twocf:registration timestamp=1401380832939, value=republican
The following query from drill fails due to empty values instead of null:
select cast(row_key as integer) voter_id, convert_from(onecf['name'], 'UTF8') name, cast(twocf['age'] as integer) age, cast(twocf['registration'] as varchar(20)) registration, cast(threecf['contributions'] as decimal(6,2)) contributions, cast(threecf['voterzone'] as integer) voterzone,cast(fourcf['create_date'] as timestamp) create_date from voterspaces where row_key < 20;
message: "Failure while running fragment. < NumberFormatException:[ ]"
From hive, running the same query against a hive table that is mapped to the same hbase table succeeded. If this how it expected to work in drill, then we should to document it.