Is there any way in TASM to set up an exception to abort queries based on the time in which they are in "acive state"?
For example, if a query is in delayed state for half an hour and then it goes into active state (the total time on system is half hour but I would consider this the start time and if the start time crosses 1 hour, the query should abort)
If I put the exception on "Duration" that time would include (time in active state + time in delayed state) and this is not working for me.
Can you please let me know if exception can be done on "active state" time?
Have you tried the "Elapsed TIme" in the exception criteria? I believe that should give you the time since the query came into the active state.
The "Elapsed Time" criteria is the total time which the query was present on the system. Like I said, it is the sum of the time the query was in delayed state because of TASM settings and the time the query was actually processed by CPU.
What I want is to be able to place an exception on the time, the query was actually processed and not on the total time the query was present on the system.
In 15.10, you will have two new TASM options when you add a new exception to a workload in Workload Designer and the exception criteria is Elapsed Time. The first option is to exclude blocked time, and the second is exclude throttle delay time. So once you get on 15.10 you will be able to do what you describe above.
Currently, there is no way to do that as a TASM exception.
Whenever any SQL session at database is not responding since long(1 hr), then it is recommended to kill that session at database level.
In which scenario the db session is hanged and required DB to restart?
i reuse that old subject to ask about CPU seconds in qualification time for TASM exceptions :
what is it made of ? how to compare with AMPCPUTIME and DELAYTIME and SEQRESPTIME in DBQLOGTBL ?
TASM guide (15.00) refers to "the accumulated CPU qualification seconds" :
does it mean i have to fulfill qualification time with an estimated number of seconds through all CPUs (let's say 24 on our system) ?
what if that number of CPU seconds is too high to be relevant ?
Happy new year everybody,