[MDEV-11402] Running connections hang forever when all tables are flushed and locked for a few seconds Created: 2016-11-29 Updated: 2016-11-29 Resolved: 2016-11-29 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Storage Engine - InnoDB |
| Affects Version/s: | 10.1.19 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Major |
| Reporter: | Michael Graf | Assignee: | Unassigned |
| Resolution: | Not a Bug | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Windows Server 2008 R2 Enterprise |
||
| Attachments: |
|
| Description |
|
For the backup of MariaDB we use VSShadow and therefore the following commands are executed: DROP TABLE IF EXISTS ibbackup_binlog_marker; With MariaDB 5.5 we didn't have any problems. But with MariaDB 10.1 all running connections hang forever and didn't return anymore when the commands above are executed. I have attachted the response of "show full processlist" and the minidump of MariaDB and the my.ini. |
| Comments |
| Comment by Vladislav Vaintroub [ 2016-11-29 ] |
|
I think you need "UNLOCK TABLES" after FLUSH TABLES WITH READ LOCK. |
| Comment by Elena Stepanova [ 2016-11-29 ] |
|
As wlad said above, you need to unlock tables after FLUSH TABLES WITH READ LOCK. The whole point of running it is to block updates from other connections, which is what happens in your processlist. 5.5 is no different in this regard, I've just checked. Maybe your flow was somewhat different with 5.5. |
| Comment by Michael Graf [ 2016-11-29 ] |
|
Sorry, it's our mistake. |