Rahul Challapalli created DRILL-1623:
----------------------------------------
Summary: COUNT (*) query with a filter does not work over
partitioned data in text and json formats
Key: DRILL-1623
URL: https://issues.apache.org/jira/browse/DRILL-1623
Project: Apache Drill
Issue Type: Bug
Reporter: Rahul Challapalli
Attachments: json.error, json.plan, lineitem_text.zip,
orders_json.zip, text.error, text.plan
git.commit.id.abbrev=6dca24a
The below 2 queries fail. The underlying data is partitioned and is one level
deep.
Text Query :
{code}
select count(*) from lineitem_text where dir0=1993;
Query failed: Failure while running fragment. Readers needs at least a column
to read. [499ab624-616a-409a-a93c-c1dc9a0d31b1]
Error: exception while executing query: Failure while trying to get next result
batch. (state=,code=0)
{code}
Json Query :
{code}
select count(*) from orders_json where dir0=1993
Query failed: Failure while running fragment. json record reader requires at
least a column [dc7cfa54-6d6f-4bf3-a0f2-38da9e58ace9]
Error: exception while executing query: Failure while trying to get next result
batch. (state=,code=0)
{code}
When I tested this against parquet and hive they seem to be having no issues.
Attached the data, logs and plans for the above 2 queries
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
