Details
-
Bug
-
Status: Resolved
-
Blocker
-
Resolution: Fixed
-
Impala 2.0
-
None
-
None
-
commit e4003483740e8fa5fafea7747b9ce3b6c1cfd9a7
Author: Victor Bittorf <victor.bittorf@cloudera.com>
Date: Wed Oct 1 21:32:31 2014 -0700
IMPALA-1337: Aggregation failures for VARCHAR
on top of
commit 27097e4e0d0ace2fd5488700d484eeb8122c2f0d
Author: Victor Bittorf <victor.bittorf@cloudera.com>
Date: Wed Oct 1 14:45:04 2014 -0700
IMPALA-1322: Fix related issuecommit e4003483740e8fa5fafea7747b9ce3b6c1cfd9a7 Author: Victor Bittorf < victor.bittorf@cloudera.com > Date: Wed Oct 1 21:32:31 2014 -0700 IMPALA-1337 : Aggregation failures for VARCHAR on top of commit 27097e4e0d0ace2fd5488700d484eeb8122c2f0d Author: Victor Bittorf < victor.bittorf@cloudera.com > Date: Wed Oct 1 14:45:04 2014 -0700 IMPALA-1322 : Fix related issue
Description
This is using the patch to fix IMPALA-1337.
After GROUP BY char_col there may be multiple NULL values in the grouping column result set.
[localhost.localdomain:21000] > create table text_types as select cast(id as char(127)) char127_col, cast(int_col as char(129)) char129_col, cast(bigint_col as varchar(256)) varchar256_col, cast(tinyint_col as string) string_col from alltypesagg; Query: create table text_types as select cast(id as char(127)) char127_col, cast(int_col as char(129)) char129_col, cast(bigint_col as varchar(256)) varchar256_col, cast(tinyint_col as string) string_col from alltypesagg +-----------------------+ | summary | +-----------------------+ | Inserted 11000 row(s) | +-----------------------+ Fetched 1 row(s) in 0.91s [localhost.localdomain:21000] > create table text_types2 as select cast(bigint_col as char(127)) char127_col, cast(tinyint_col as char(129)) char129_col, cast(int_col as varchar(256)) varchar256_col, cast(id as string) string_col from alltypesagg; Query: create table text_types2 as select cast(bigint_col as char(127)) char127_col, cast(tinyint_col as char(129)) char129_col, cast(int_col as varchar(256)) varchar256_col, cast(id as string) string_col from alltypesagg +-----------------------+ | summary | +-----------------------+ | Inserted 11000 row(s) | +-----------------------+ Fetched 1 row(s) in 0.90s [localhost.localdomain:21000] > create table foo as select t2.char127_col AS char127_col_1 FROM text_types t1 LEFT JOIN text_types t2 ON t2.char127_col = t1.char129_col GROUP BY t2.char127_col; Query: create table foo as select t2.char127_col AS char127_col_1 FROM text_types t1 LEFT JOIN text_types t2 ON t2.char127_col = t1.char129_col GROUP BY t2.char127_col +----------------------+ | summary | +----------------------+ | Inserted 1019 row(s) | +----------------------+ Fetched 1 row(s) in 69.77s [localhost.localdomain:21000] > select * from foo order by char127_col_1 desc limit 10; Query: select * from foo order by char127_col_1 desc limit 10 +---------------+ | char127_col_1 | +---------------+ | NULL | | NULL | | NULL | | NULL | | NULL | | NULL | | NULL | | NULL | | NULL | | NULL | +---------------+ Fetched 10 row(s) in 0.30s
It's also interesting that the CTAS took 70 seconds. I'll look into that a little.