[DRILL-2801] ORDER BY produces extra records

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Duplicate
• Affects Version/s: 0.8.0
• Fix Version/s: 1.0.0
• Component/s: Execution - Relational Operators
• Assignee: Steven Phillips
• Reporter: Sudheesh Katkam
• Created: 15/Apr/15 19:11
• Updated: 28/Apr/15 18:27
• Resolved: 28/Apr/15 18:27

Description

Running in embedded mode on my mac.

$ wc -w data.csv
   50000 data.csv

Here's the query:

0: jdbc:drill:zk=local> SELECT count(*) FROM dfs.`data.csv`;
+------------+
|   EXPR$0   |
+------------+
| 50000      |
+------------+
1 row selected (0.223 seconds)
0: jdbc:drill:zk=local> SELECT columns[0] FROM dfs.`data.csv` ORDER BY columns[0];
+------------+
|   EXPR$0   |
+------------+
...
| 6          |
+------------+
50,001 rows selected (0.928 seconds)
0: jdbc:drill:zk=local> SELECT tab.col, COUNT(tab.col) FROM (SELECT columns[0] col FROM dfs.`data.csv` ORDER BY columns[0]) tab GROUP BY tab.col;
+------------+------------+
|     col      |   EXPR$1   |
+------------+------------+
| 2          | 10000      |
| 3          | 10000      |
| 4          | 10000      |
| 5          | 10001      |
| 6          | 10000      |
+------------+------------+
5 rows selected (0.704 seconds)

Issue Links

• duplicates: DRILL-2083 order by on large dataset returns wrong results