We thought of that -- but updating the clients would be tough to do, since we have a lot of 3rd party apps hitting the box. =20 We took one busy database out of the loop, and that improved things slightly,=20 ping times in the 40K now, where before the tnsping just sat there forever,=20 and now you can get in in about 30-40 seconds, compared to up to 15 minutes before: [03-FEB-2005 10:01:00:480] nscon: doing connect handshake... [03-FEB-2005 10:01:00:480] nscon: sending NSPTCN packet [03-FEB-2005 10:01:36:705] nscon: got NSPTRS packet [03-FEB-2005 10:01:36:706] nscon: sending NSPTCN packet =20 -----Original Message----- From: Hemant K Chitale [mailto:hkchital@xxxxxxxxxxxxxx]=20 Sent: Thursday, February 03, 2005 9:35 AM To: Thomas Jeff; oracle-l@xxxxxxxxxxxxx Subject: RE: 9.2.0.4 listener problems Curious. Have you tried setting the listener to listen on another port ? Setup the listener.ora to use another port [say 1523 or 1621] other than=20 the default 1521. {of course, you'd have to update the tnsnames.ora on the clients as well -- or you could test this with two seperate listeners on the two seperate ports} Hemant At 09:20 PM Thursday, Thomas Jeff wrote: >To add to this. We shut down the 9i listener and fired up a 10g >listener. The same >problems persist. What happens is that it's fine for about 5 minutes >(with the 10g >listener) then the tnsping times drop off the table, and the slowdowns >resume. It >looks like some sort of load/resource contention issue. > Hemant K Chitale http://web.singnet.com.sg/~hkchital =20 -- //www.freelists.org/webpage/oracle-l