Oracle Parallel Client Failover (Server down)

From: James Gardner <nomail.sr34_at_dial.pipex.com>
Date: Sun, 5 Sep 1999 18:26:57 +0100
Message-ID: <7qu8t9$98m$1@lure.pipex.net>

I have installed Oracle 8.0.5.1 in a two node concurrent cluster on AIX 4.3.2 (HACMP 4.3). Unfortunately this install isn't on an IBM SP which would solve a lot of the problems.

I have the pc clients failing over via SQL*NET using preconnects (specified in the tnsnames.ora) and MTS for load balanacing (as specified in the OPS Admin and Config manual, chapter 22).

In the event that the database is down on a single node clients failover in a matter of seconds. However if a server is down ie. there is no reponse from the ip address the client failover takes upwards of 2 minutes plus which is unacceptable to my customer. Has anybody encountered this problem if so what the hell is the problem because I can spot it.

Regards

James Gardner Received on Sun Sep 05 1999 - 12:26:57 CDT