[IMPALA-5286] Query with Kudu col name w/ different casing from 'order by' fails

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 2.8.0
• Fix Version/s: Impala 2.10.0
• Component/s: Frontend
• Labels: kudu
• Epic Color: ghx-label-6
• Assignee: Thomas Tauber-Marshall
• Reporter: Matthew Jacobs
• Created: 05/May/17 20:50
• Updated: 19/Jun/17 18:22
• Resolved: 19/Jun/17 18:22

Description

When a Kudu table is created outside Impala with non-lower-case column names, an Impala query using that column name in 'order by' will fail.

Example:

Create a Kudu table in the python client:

import kudu
builder = kudu.schema_builder()
builder.add_column('key').type(kudu.int64).nullable(False).primary_key()
builder.add_column('VAL', type_=kudu.int64, nullable=False)
schema = builder.build()
from kudu.client import Partitioning
partitioning = Partitioning().add_hash_partitions(column_names=['key'], num_buckets=3)
c.create_table('foo', schema, partitioning)

Then in Impala:

create external table foo stored as kudu TBLPROPERTIES ( 'kudu.table_name' = 'foo' );

select * from foo order by VAL;

results in the failure: Unable to deserialize scan token: Invalid argument: Duplicate column name: VAL

While Impala cannot support different cased column names that conflict (e.g. "VAL" and "val" should not be supported, see IMPALA-5283), this query could be supported.

Issue Links

• relates to: IMPALA-5283 Handle case sensitivity naming conflicts in Kudu tables