Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Duplicate
-
10.2.6
-
Windows Server 2012 R2 Standard
Description
After the update from MariaDB 5.5.x to 10.2.6 the startup of a mirrored system takes now 5 - 50 minutes - with MariaDB 5.5.x the startup took not longer than 10 - 30s.
The MariaDB server was stopped always regularly - so no recovery action is necessary.
The cause for this behavior seems to be the page_cleaner.
Here is an extraction from the MariaDB errorlog:
2017-08-21 13:42:00 7676 [Note] InnoDB: Highest supported file format is Barracuda.
2017-08-21 14:32:36 7676 [Note] InnoDB: 128 out of 128 rollback segments are active.
2017-08-21 14:32:36 7676 [Note] InnoDB: Creating shared tablespace for temporary tables
2017-08-21 14:32:36 7676 [Note] InnoDB: Setting file '.\ibtmp1' size to 12 MB. Physically writing the file full; Please wait ...
2017-08-21 14:32:36 7676 [Note] InnoDB: File '.\ibtmp1' size is now 12 MB.
2017-08-21 14:32:36 7676 [Note] InnoDB: Waiting for purge to start
2017-08-21 14:32:36 7676 [Note] InnoDB: 5.7.14 started; log sequence number 9834561730508
2017-08-21 14:32:36 19676 [Note] InnoDB: page_cleaner: 1000ms intended loop took 3036315ms. The settings might not be optimal. (flushed=0 and evicted=0, during the time.)
Attachments
Issue Links
- relates to
-
MDEV-13941 Innodb/Windows, 10.2 : High NTFS fragmentation
- Closed