Re: TCP Queuing Problem

From: foo <foo_at_bar.com>
Date: 1996/11/18
Message-ID: <329146AD.4E01_at_bar.com>#1/1


Thanks for those of you who suggested lsof for mapping ports to sockets. It's definitely easier than using crash.

We were able to trace the problem. It seems that the Tuxedo servers that were our Oracle clients were idle most of the time. Apparently SQL*Net v2 sends keepalive messages that are acknowledged by the client side code whenever an SQL call is made. No sql call, no processing of the TCP queues, hence socket level queuing because no one is reading the data. The answer seems to be shutting down unused Oracle clients (we were piloting a new release that wasn't really being executed). Received on Mon Nov 18 1996 - 00:00:00 CET

Original text of this message