04-07-2009 11:30 AM
hello,
I have setup a job and its been running since 1 day and dont know why it has taken such a long time.
The job is simply writing data on the new created info structure which is transactional data.
So I am not sure why it is taking so long.
How can I make some checks etc..
How should I do the Trace and steps etc..
Please help
04-07-2009 12:19 PM
wriiting means writing to database?
If yes, then you can start the SQL Trace in parallel, just activate for 10 or 20 min and analyse it, maybe
one process is extremely slow.
Otherwise check STAD
And SM50
Siegfried
04-07-2009 11:48 AM
use the transactions SM50 and Sm37 to analyse
Regards,
ravi
04-07-2009 12:04 PM
Hi Andy,
Maybe you are facing problems with:
- Nested loops (loop inside loop w/o proper treatment)
- Read non indexed internal tables
- SELECT SINGLE, inside Loops
You can start your analysis with SE30, put the analyser running in parallel with your job.
Go to SE30
-> Switch On (In Parallel Session)
-> Select your process (just one click)
-> Start measurement (Shift+F7)
-> At this point trace data will be collected, wait at least one minute here
..
-> Deactivate the trace click on End Measurement (Shift+F1)
-> Back to first screen
-> Click on button "Other Files..."
-> Just go ahead (selecting your user)
-> The measurement file is shown, double click it
-> A graphic comparing ABAP time and DB time is shown
-> Go to the hit list to view the codes that is taking time
-> This will show you in what step of your execution the program is taking time
The perfect option is use SE30 in all your job execution, unfortunately it will generate a lot of information
Start the optimization from the beggining, removing all bottlenecks you find through the way
Check this, from Sigfried:
[https://www.sdn.sap.com/irj/sdn/weblogs?blog=/pub/wlg/7769] [original link is broken] [original link is broken] [original link is broken];
Tip: If I were you I would start the investigation from the beggining with few data, just to understand the behaviour of the job and identify the possible bottlenecks.
Regards,
Fernando Da Ros
04-07-2009 12:19 PM
wriiting means writing to database?
If yes, then you can start the SQL Trace in parallel, just activate for 10 or 20 min and analyse it, maybe
one process is extremely slow.
Otherwise check STAD
And SM50
Siegfried
04-07-2009 12:41 PM
hello,
how do I execute the trace?
Can you explain me the steps?
thanks
04-07-2009 12:53 PM
hi,
for sql trace use TCODE ST05 and
select SQLtrace check box and
select activate the trace
run ur program and
deactivate trace .
select display trace .
SQL trace:
/people/siegfried.boes/blog/2007/09/05/the-sql-trace-st05-150-quick-and-easy
~linganna
Edited by: katigiri linganna on Apr 7, 2009 1:54 PM
04-07-2009 8:33 PM
hello,
I get a graph based on ABAP programme, DB table and R/3 and it shows that is running.
The ABAP programme shows green sign and other shows red. What does that mean?
Thanks
04-07-2009 8:45 PM
Press the "Hit list" button (F5) on the graph screen. That will tell you where the time is spent.
Rob
04-07-2009 8:46 PM
Sm50 > in one of the line it will show the program name or the user who executed the program, hit the check box on that row and
progam/ mode > program > debugging and switch the debugger on and see where its taking some much of time...
be care full when you exit debugger.
04-08-2009 10:32 AM
you don't get a graph with the SQL Trace only with the ABAP-Trace, but the ABAP Trace is useless for such a long running process.
Check the STAD to get an overview, and hope that the are high database times, which can be addressed by the SQL Trace.
There are two blogs by me on the two traces, which were already mentioned. The SE30 blogs tells you ecactly what the colors of the SE30 mean ... more or less nothing.
Siegfried
04-08-2009 1:48 PM
I am working on 4.0B and the system doesnt understand the transaction...STAD
04-08-2009 1:55 PM
Hi Andy,
on 4.0B use transaction STAT
Kind regards,
Hermann
04-08-2009 1:57 PM
is it transaction or programme? can you give me the programme name if possible?
04-08-2009 2:01 PM
Hi Andy,
STAT is the transaction.
I don't have a 4.0B here anymore. But you can look up the program name
yourself in transaction SE93. - Enter the Transaction - display... see filed Program.
Kind regards,
Hermann
04-08-2009 2:08 PM
Thanks, the job is running though but with STAT I dont see any statistical records..
04-08-2009 2:29 PM
Hi Andy,
yes... you will see the statistical record after
- the program has finished or
- the program is aborted (not on 4.0B but in new releases ...).
While it is running... you can only use
- SM50 (refresh several times) - to see what the program is doing NOW
- SM50 (doubleclick) - to get some acummulated values with resp. to DB accesses
- SE30 / ST05 / ST12 - to trace the current execution
- Debugger - to get the call stack and current memory consumption
Kind regards,
Hermann
04-14-2009 9:35 AM