I am loading data via SQL*Loader on Oracle 8.1.6.0.0. There are up to 5
sqlldr sessions running at once. This has been going well for a number
of weeks. The client and server boxes (both Solaris 2.7) were moved to
new location (new IP addresses) and since then the loading has slowed
considerably, the main reason being'SQL*Net break/reset to client' wait
events. A good indication of the degradation is the redo logs switching
every 2.5 hours instead of the previous half an hour. I have looked
this up on Metalink and the general opinion seems to be that it is not
a problem . However, STATSPACK reports show that this wait event
constitutes a staggering 70% to 92% of total waits. There does not seem
to be anything amiss in the listener.log or alert.log. I have no
trouble connecting to the server from the client via SQL*Plus, or
pinging each box from the other.
The only change made in the Oracle configuration was the hostname in
the *.ora files, which I have since replaced with the IP address, to no
avail.
My questions is:
Can I confidently say that this is not an Oracle, but a network problem?
--
Gerry
Sent via Deja.com http://www.deja.com/
Before you buy.
Received on Sat Nov 04 2000 - 03:34:36 CST
