on 01-04-2013 4:02 AM
Hello everyone,
I'm doing a system monitoring job.
And find the roll wait time counts for 30% of the total response time.
attachment is the details.
My question is :
How to analyse the problem according to this situation?
Is it a network pb btw Front end to the application server?
So my first step is to niping these 2 ends?
Thank you for show me the guideline or road map to analyse it.
Best regards,
Kate
Hello Kate,
The point of interest in these cases is the "Parts of Response Time" tab in your second screenshot.
What % of Dialog Response time is consumed in Roll Wait Time ? I know you have mentioned 30% still paste the screenshot.
If it is very less, we need not bother.
However to anwer your primary query, you need to enable RFC trace for user/program/work process/job etc if you want to troubleshoot HIgh Roll Wait time.
If GUI Time is also more along with Roll Wait time for your Dialog Response, then it can be a issue with Front End to App Server connectvity. NIPING can give you an idea, you need to identify few terminals.
Thanks,
Samik
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello Samik,
Thank you very much for your answer.
Here i attached more details pic about the 2 Tcode that consume the high roll wait and GUI time today.
So for this trascation all the user are the same hign consume
So what am i supposed to do ?
I want to do the Front end to App server NIPING
but how to locate the terminal since i just have the user id.(who should i turn to?)
BR,
Kate
Hello Kate,
Looks like for those tcodes all experiencing high GUI times. You can launch userid/terminal search hunt with the help of functional folks responsible for these applications(tcodes), for your own troubleshooting you can start ST05 RFC traces for those tcodes for a limited amount of time and analyze the output.
Also since these are Z tcodes, we need to see at runtime with the user what it actually does, whether they are used simply to download huge data for reporting etc, because if that is the case Gui time will increase. So you got to identify 1 user and look how they are running hte transactions.
Thanks,
Samik
Hello Samik,
Thank you for your quick reply, and sorry for the delay, busy with other urgent work.
Here i still have some questions:
1."userid/terminal search hunt with the help of functional folks responsible for these applications(tcodes)"
for this i don't quite understand, do you mean that i should ask the fonction responsible to NIPING the 2 ends? if not , could you please explain more to me?
2. for the ST05 , will it make huge work load when i do it on the production system? Just do a short period around 15 min won't hurt much right?
3. I wonder if it is also practical to create a user for me to test this transaction ? what do you think?
Thank you very very much~~
BR,
Kate
Hi Kate,
1. Functional will not be able to run NIPING, you have to run it. But first you need to identify certain users who are actually using those transactions (The nos doesn't really help much). You got to see it in live (or screen sharing) what they are doing with the transaction.
2.As a proactive measure you can start ST05 tracing for those 2 tcodes. 15 mins is OK, doesn't hurt much.
3.Yes, if you can run the problem tcode it is good, but you have to run it the way the users are actually using it. In production it is better to have the functional team run those transactions for you and you watch or trace.
Thanks,
Samik
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.