[DRILL-4898] wrong results : Query on directory containing CSV data

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Not A Problem
• Affects Version/s: 1.9.0
• Component/s: Execution - Flow

Environment

4 node cluster

• Assignee: Unassigned
• Reporter: Khurram Faraaz
• Created: 21/Sep/16 10:24
• Updated: 22/Sep/16 18:35
• Resolved: 22/Sep/16 18:35

Description

incorrect results : Query on directory containing CSV data
directory has 4534327 number of rows ~ 4.5M records (there are 6 CSV files)
Drill 1.9.0 commit ID: f3c26e34
I can share the data to reproduce the issue.

Note that data in columns[3] has the value "B02512\r" in query results.

 
0: jdbc:drill:schema=dfs.tmp> select * from `uber_trip_data` limit 5;
+----------------------------------------------------------+
|                         columns                          |
+----------------------------------------------------------+
| ["2014-08-01 00:03:00","40.7366","-73.9906","B02512\r"]  |
| ["2014-08-01 00:09:00","40.726","-73.9918","B02512\r"]   |
| ["2014-08-01 00:12:00","40.7209","-74.0507","B02512\r"]  |
| ["2014-08-01 00:12:00","40.7387","-73.9856","B02512\r"]  |
| ["2014-08-01 00:12:00","40.7323","-74.0077","B02512\r"]  |
+----------------------------------------------------------+
5 rows selected (0.184 seconds)

But when we do a select on columns[3] we see a different value in the query result.

0: jdbc:drill:schema=dfs.tmp> select columns[3] from `uber_trip_data` limit 5;
+----------+
|  EXPR$0  |
+----------+
  |02512
  |02512
  |02512
  |02512
  |02512
+----------+
5 rows selected (0.159 seconds)  

Searching for 'B02512' returns no rows. (where as it should have returned data)

0: jdbc:drill:schema=dfs.tmp> select * from `uber_trip_data` where columns[3]='B02512';
+----------+
| columns  |
+----------+
+----------+
No rows selected (1.707 seconds)