6 Replies Latest reply on May 23, 2017 9:49 AM by 3209374

    Measuring EBS R12.2 session Idle time



      trying to balance security rules and user efficiency in EBS R12.2



      In EBS the standard Oracle time out session is set ot 30 minutes. However my users and our internal processes are so that it can be that users are idle in EBS R12 longer than the 30 minutes.

      With the standard setting they then need to login again to continue their session. This is leading to severe inefficiency and complains from the users.

      To come to an optimum I would like to measure (and store) for a (specific) group of user the avg session idle time = (so how long are they inactive in a session before getting back to EBS R12 and continue).

      As Oracle EBS is able to measure the idle time (otherwise how could it log off the user after the session idle time parameter ie reached?) I believe there must be a way to grab this info and store it in a table / log.

      I do believe Oracle is just starting and resetting the session idle timer per user and not storing this so I need a way to write this into a table / log



      Is there anybody that has created a measurement script for this already?

      Are there any suggestions / ideas on how to do this best.

        • 1. Re: Measuring EBS R12.2 session Idle time

          There are 3 settings which control session timeout in R12 including 12.2:


          Site Level Profile Options:


          ICX: Limit Time --hours

          ICX: Session Timeout  --minutes


          Context Variable:


          <session_timeout oa_var="s_sesstimeout">1800000</session_timeout>  --milliseconds


          The defaults are 4 hours for ICX: Limit Time, 30 minutes for ICX: Session Timeout  and 1800000 milliseconds for s_sesstimeout.  The value for s_sesstimeout in milliseconds must match ICX: Session Timeout.


          The ICX: Limit Time will terminate a session after the limit is reached no matter what they are doing.  The user could be heads down entering transactions all morning and at 4 hours will be terminated.  This is not ideal behavior so I recommend increasing ICX: Limit Time to 10 or 12 hours to at least allow users to get through a shift.


          The session timeout of 30 minutes in my opinion is too low because users get terminated for going to a quick meeting.  If I have to close my browsers and log back into production 8-10 times a day it is a huge drag on efficiency as you describe.  The infrastructure cost of having more idle sessions sitting on the web and database server is much less than the annoyance to users forced to log back into Oracle all day long to do their job.  I set the Session Timeout to anywhere between 1 hour and 4 hours, and usually 10 hours in non prod clones.


          Then ICX: Session Timeout => 600

          <session_timeout oa_var="s_sesstimeout">36000000</session_timeout>


          This recommendation will likely be an outlier from what others are doing but I feel IT shouldn't be an impediment to getting work done.  It is batch jobs, active forms sessions and runaway frmweb processes which consume resources anyway.  Having a low session timeout will not prevent runaway frmweb sessions from using 100% CPU either.

          • 2. Re: Measuring EBS R12.2 session Idle time
            Pravin Takpire


            i think you can check FND_LOGINS, FND_SESSIONS and v$session. I think combination of these 3 tables might throw some light. In my knowledge there is no clear cut way/tables where it keeps session_idle information.




            • 3. Re: Measuring EBS R12.2 session Idle time

              Hi mdtaylor, thanks for your reply.

              The only point that I would like is to see if the session idle time can be measured vs just setting it to a arbitrary number. But your feedback for sure is valuable pointing out the various settings

              best regards,


              • 4. Re: Measuring EBS R12.2 session Idle time

                Hi Pravin,


                will check these areas, and if it contains the information I am looking for, will see if we can create a program that can store the information for selected users during a framed period of time so hat I can change the session idle time to meet business expectation



                • 5. Re: Measuring EBS R12.2 session Idle time

                  Hi Marc,


                  Session idle time trackers are client side APIs that run in the local browsers and java applets.  Oracle is not sending each wait or idle state back into a table on the database server.  If it did, large enterprises would bog down as thousands of users hit the system and wait events started pouring back into the database.  If anything was written back, it would be start of session and end of timeout, which would just be 30 minutes after last activity if you don't change the timeout.  If you want to measure this, you would have to create some logic to track it and record it.  It would be easier to do this in CUSTOM.pll rather than OAF pages most likely.




                  • 6. Re: Measuring EBS R12.2 session Idle time

                    Hi Michael, thanks for this I will look into your suggestion.