SQL*Net expire_time parameter

I am trying to figure out how the SQL*Net V2 parameter SQLNET.EXPIRE_TIME works. I know that I can set a value in minutes for it, and that it uses tcp's keepalive to check for a connection and keep or kill the connection.

What I don't know, and haven't seen explained clearly, is this; If my PC crashes and I reboot and reconnect within the time of my expire_time parameter, Will SQL*Net know that my PC went away? In other words, does it check using session or process id, or simply using ip address?

I am hoping it is the former, since we need to be able to kill dropped connections which may have row-level locks on them from SELECT... FOR UPDATE statements. If it is the latter, people would have to stay completely off until after the expire_time passes.

Any input on this gratefully accepted.

--
Bob Blizard    rblizard_at_cmcsys.com     | All these worthwhile
CMC Systems, Inc.                      | opinions are mine, and
175 Littleton Rd   Westford, MA 01886  | not necessarily CMC's...
Ph  (508) 392-1300 Fax (508) 392-1303  |
"If you want to park in the first row of life, 
            you must go there to look for a space."