[MDEV-10403] Garbage characters in show processlist with ANALYZE TABLE on slave Created: 2016-07-20  Updated: 2016-11-24  Resolved: 2016-11-24

Status: Closed
Project: MariaDB Server
Component/s: Storage Engine - TokuDB
Affects Version/s: 10.1.16
Fix Version/s: 10.1.18

Type: Bug Priority: Major
Reporter: jocelyn fournier Assignee: Unassigned
Resolution: Fixed Votes: 0
Labels: upstream


 Description   

Hi,

From time to time, on a slave server, I notice the replicated ANALYZE TABLE statement is displayed with garbage in the State column during a SHOW PROCESSLIST :

e.g. : 

MariaDB [(none)]> show processlist;
 
+------+-------------+-----------+------+---------+-------+----------------------------------+-------------------------------------------+----------+
 
| Id   | User        | Host      | db   | Command | Time  | State                            | Info                                      | Progress |
 
+------+-------------+-----------+------+---------+-------+----------------------------------+-------------------------------------------+----------+
 
|   37 | system user |           | NULL | Connect | 48841 | Waiting for master to send event | NULL                                      |    0.000 |
 
| 1191 | root        | localhost | sc_2 | Sleep   |  2789 |                                  | NULL                                      |    0.000 |
 
| 1246 | system user |           | sc_2 | Connect |     3 | @R
 
??                             | ANALYZE TABLE `sc_stat_product_type_user` |    0.000 |
 
| 1336 | root        | localhost | NULL | Query   |     0 | init                             | show processlist                          |    0.000 |
 
+------+-------------+-----------+------+---------+-------+----------------------------------+-------------------------------------------+----------+
 
4 rows in set (0.00 sec)
 
[...]
 
 
 
| 49 | system user |           | sc_2 | Connect |    5 | ???T?                               | ANALYZE TABLE `sc_stat_product_type_user` |    0.000 |

Note sc_stat_product_type_user is a TokuDB table (so perhaps the bug is upstream ?)

Thanks and regards,
Jocelyn Fournier

 Comments   
Comment by Elena Stepanova [ 2016-07-20 ]

I'm not sure it's upstream. We had a similar issue before, regardless TokuDB: MDEV-4723. It was fixed years ago, but maybe it's re-appeared. I'll see if I can reproduce it.

Comment by Elena Stepanova [ 2016-07-21 ]

It is upstream after all. I can reproduce it very easily with a stress test, both on MariaDB 10.0-10.2 and Percona Server 5.6, but not on Percona 5.7, so I guess it's been fixed and there is no point filing it upstream.
It's not reproducible anywhere with XtraDB, so it's TokuDB-specific.

serg, do we want to fix it in our trees or backport the bugfix from Percona 5.7? If we do, I'll try to make something more deterministic than my stress test, otherwise there is no point.

Comment by Rich Prohaska [ 2016-07-21 ]

The changes in https://tokutek.atlassian.net/projects/DB/issues/DB-1003 may fix this problem. The analyze code is setting proc info to the address of a local variable that may go out of scope.

Comment by jocelyn fournier [ 2016-11-24 ]

This one seems to be fixed now

Comment by Elena Stepanova [ 2016-11-24 ]

joce,

Thanks for update.
Closing as fixed in 10.1.18, because that's the last release when TokuDB merge happened (5.6.32-78.1).

Generated at Thu Feb 08 07:41:58 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.