Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.1.16
-
This problem was detected in a CentOS 7 and Fedora 23 environment
Description
After the upgrade from the 10.1.14 to 10.1.16, I have detected that some queries need a lot of time to be executed.
When I have downgraded to the 10.1.14, I have checked that the problem has dissapeared. I have executed the mysql_upgrade after each upgrade.
I attach some files that I expect that help you to understand the problem and fix it.
- saltos-10.1.14.log => contains the log of all queries executed by SaltOS to get the list page - using a MariaDB 10.1.14.
- saltos-10.1.16.log => contains the log of all queries executed by SaltOS to get the list page using a MariaDB 10.1.16.
- query.sql => contains the conflictive query!!!
- explain-10.1.14.log => contains the explain extended command for the previous query using a MariaDB 10.1.14.
- explain-10.1.16.log => contains the explain extended command for the same query using a MariaDB 10.1.16.
- schema.sql => contains the schema used by the SaltOS's database.
- used-size.pdf => contains the real used size to understand the problem.
Ok, the change in behavior that I'm observing was caused by the fix for
MDEV-10006.It also matches the behavior that sanz is reporting:
was the only possible.
should be possible also, this is what
MDEV-10006is about)MDEV-10006made the {e, a}order possible
(now, looking if there is anything we could do here)