Re: idle_time resource limit

An interesting feature of the idle_time resource limit is that the user's session is not terminated until Oracle is able to notify the user that he has exceeded the idle_time limit. For the notification to take place the user must initiate communication to the instance. At that time Oracle will respond with a message indicating that idle time has been exceeded. Only after Oracle has issued that message will pmon come in and clean up the resources held by that user.

An alternative method that I have seen used to kick off idle users is to schedule a script to periodically query the database finding sessions with excessive idle times, getting the operating system process id from v$session and then killing the session from the operating system side. Eventually pmon will wake up and notice that the operating system process is gone and will clean up the session in Oracle.

Hope this helps,

Mary Travis

"Carol Roy" <royc_at_ci.portsmouth.va.us> wrote:

>I tried a brief test setting idle_time to 30 and leaving a user in a form
>accessing rows.
>After at least an hour, the user session was still showing up, there were
>locks on the rows and when I tried to kill the user, I got a message that
>said something like session does not exist.
 

>Anyone out there using this option. Have you found it works well?

Mary Travis
BellSouth Telecommunications
email: Mary.B.Travis_at_bridge.bellsouth.com Received on Tue May 13 1997 - 00:00:00 CDT