We have a CAS with a "slow" database. Therefore we noticed our weekly db maintenance tasks are failing within the default time. We increased the timeout step-by-step and are now at 13h timeout. We noticed the following log messages:
T DB 18-04-01 01:00:02.589 Checking defragmentation statistics ...
T DB 18-04-01 01:13:13.393 Defragmentation statistics collected. Duration: 13:10.804.
T DB 18-04-01 01:13:34.917 Defragmentation. 1 of 73 done. Duration: 00:20.321, Delay increase 00:20.321
T DB 18-04-01 11:48:18.559 Defragmentation. 32 of 73 done. Duration: 10h 35:03.963, Delay increase 0-8:0-14.00-596
T DB 18-04-01 11:48:18.559 Leaving idx defragmentation. No time to defragment 41 tables...
T DB 18-04-01 11:48:18.559 Defragmentation done. Duration: 10h 48:15.970
Does anyone know why the task is ending before the timeout is reached? The whole DBWeeklyMaintenance Task is starting at 3 am server time (which is equal to 1 am log time) and has a timeout of 13h as well as the timeout for the defrag command which is 13h as well.
The other thing I noticed is that there is a daily job which should to defragmentation as well but in the logs it is mentioned
1 SCHED 18-04-04 19:00:00.011 Task Server task DefragIndexesOnChangedTables is running in new thread now.
1 SCHED 18-04-04 19:00:00.011 DefragIndexesOnChangedTables Start executing command DefragIndexesOnChangedTables from server task DefragIndexesOnChangedTables for 2018-04-04 21:00
T ADM 18-04-04 19:00:00.011 DefragIndexesOnChangedTables is switched off
1 SCHED 18-04-04 19:00:00.011 DefragIndexesOnChangedTables End of successful execution (0) of command DefragIndexesOnChangedTables from server task DefragIndexesOnChangedTables for 2018-04-04 21:00
The task file tasks-100-hcbs.xml looks fine for this task, is there any other option to (de-)activate this task?
Thanks for any input on this.
Most probably an other scheduled task within 13 hours interrupted the DBWeeklyMaintenance and also you can check with the DBA in case they are taking incremental/full backup at the same time which is causing this problem and slowing down the DBWeeklyMaintenance process.
How many raw sessions do you have because taking more than 7-8 hours for the weekly task should be the maximum?
DefragIndexesOnChangedTables task is only configurable in the tasks-100-hcbs.xml
we have a virtualized environment which is somehow not performing properly (we know this).
We have only around 3M Raw Sessions on this CAS - therefore it should be faster but it might be slower due to virtualization.
I'm only wondering why the Daily Task for DefragIndexesOnChangedTables is active in the tasks-100-hcbs.xml but in the logs it shows a message with "Task is switched off". There must be another configuration possibility for this task but I never noticed another one.