[jira] [Created] (DRILL-1327) Any subsequent query may hung in sqlline after you cancel one query in the same session

["Hao Zhu (JIRA)" <ji...@apache.org>]

Hao Zhu created DRILL-1327:
------------------------------

             Summary: Any subsequent query may hung in sqlline after you cancel one query in the same session
                 Key: DRILL-1327
                 URL: https://issues.apache.org/jira/browse/DRILL-1327
             Project: Apache Drill
          Issue Type: Bug
          Components: Query Planning & Optimization
    Affects Versions: 0.4.0
         Environment: MapR 3.1.1
Drill 0.4.0
            Reporter: Hao Zhu
            Priority: Critical


This is another issue related to DRILL-1326.
If we cancel a query in sqlline, subsequent query may hung.
I can reproduce it almost every time.

How to reproduce:
1. Logon sqlline, run below SQL:
{code}
select *
 from cp.`employee.json` a, cp.`employee.json` b
 where a.position_id=b.position_id ;
{code}

When it is printing the results for about 3 seconds, type "ctrl-c" to stop it.
Then any subsequent query will hung.
For example,
{code}
| 110         | Josie Underwood | Josie      | Underwood  | 15          | Store Permanent Checker | 7          | 15            | 1914-02-02 | 1996-01-01 00:00:00.0 | 8200.0     | 101       |
| 110         | Josie Underwood | Josie      | Underwood  | 15          | Store Permanent Checker | 7          | 15            | 1914-02-02 | 1996-01-01 00:00:00.0 | 8200.0     | 101       |
| 110         | Josie Underwood | Josie      | Underwood  | 15          | Store Permanent Checker | 7          | 15            | 1914-02-02 | 1996-01-01 00:00:00.0 | 8200.0     | 101       |
| 110         | Josie Underwood | Josie      | Underwood  | 15          | Store Permanent Checker | 7          | 15            | 1914-02-02 | 1996-01-01 00:00:00.0 | 8200.0     | 101       |
+-------------+------------+------------+------------+-------------+----------------+------------+---------------+------------+------------+------------+---------------+-----------------+--+
11,744 rows selected (2.879 seconds)
0: jdbc:drill:zk=node1.xxx.com:5181,node3.>
Command canceled.
0: jdbc:drill:zk=node1.xxx.com:5181,node3.>
Command canceled.
0: jdbc:drill:zk=node1.xxx.com:5181,node3.>
0: jdbc:drill:zk=node1.xxx.com:5181,node3.> SELECT TABLE_SCHEMA, TABLE_NAME, TABLE_TYPE
. . . . . . . . . . . . . . . . . . . . . . .> FROM INFORMATION_SCHEMA.`TABLES`
. . . . . . . . . . . . . . . . . . . . . . .> ORDER BY TABLE_NAME DESC;

{code}

In the end, we have "ctrl-z" and "kill -9 %1" to exit the sqlline.
After that, the SQL become "Running Queries" forever.
We need to fix this because it looks like the sqlline is not stable.



--
This message was sent by Atlassian JIRA
(v6.2#6252)