[DRILL-2606] Casting a column from hbase to boolean after a join with parquet makes DRILL unresponsive

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 0.8.0
• Fix Version/s: 0.9.0
• Component/s: Execution - Flow
• Assignee: Venki Korukanti
• Reporter: Rahul Kumar Challapalli
• Created: 27/Mar/15 18:17
• Updated: 06/May/15 22:52
• Resolved: 15/Apr/15 18:22

Description

Small scale repro without HBase:

alter session set `planner.slice_target`=1;
select cast(cast(bool_col  as varchar(100)) as boolean) from dfs.`/Users/hadoop/Downloads/fewtypes_null.parquet` order by int_col;

git.commit.id.abbrev=4d398ed

The below query never returns and causes Sqlline to hang. DRILL also becomes unresponsive after executing this query. So I have to restart drill for sqlline to even come up.

select cast(cast(o.types.bool_col as varchar(100)) as boolean) bool_col
from dfs.`cross-sources`.`fewtypes_null.parquet` p
inner join hbase.fewtypes_null o
    on p.int_col = cast(cast(o.types.int_col as varchar(100)) as int)

This issue is similar to DRILL-2605. However in this case I did not use a view and some this causes drill/sqlline to become unresponsive where in DRILL-2605 we just get back an error.

I attached the data files, data loading script, and the error log. Let me know if you need anything else.

Issue Links

• is duplicated by: DRILL-2605 Projecting a boolean column from a hbase view after a join is causing an ExpressionParsingException