Puzzling through RAC load balancing on a service
From: <Christopher.Taylor2_at_parallon.net>
Date: Thu, 16 May 2013 02:44:10 -0500
Message-ID: <F05D8DF1FB25F44085DB74CB916678E8879E3BDFA8_at_NADCWPMSGCMS10.hca.corpad.net>
If I've set a goal on a service in 10g (10.2.0.4) RAC, but the client has LOAD_BALANCE=YES in the tnsnames file, can these cause screwy load balancing? What I have is a goal set for SHORT and SERVICE_TIME and I'm monitoring gv$servicemetric and I see that node1 has a goodness value of 9600, node2 = 4600, node3‚00.
Date: Thu, 16 May 2013 02:44:10 -0500
Message-ID: <F05D8DF1FB25F44085DB74CB916678E8879E3BDFA8_at_NADCWPMSGCMS10.hca.corpad.net>
If I've set a goal on a service in 10g (10.2.0.4) RAC, but the client has LOAD_BALANCE=YES in the tnsnames file, can these cause screwy load balancing? What I have is a goal set for SHORT and SERVICE_TIME and I'm monitoring gv$servicemetric and I see that node1 has a goodness value of 9600, node2 = 4600, node3‚00.
Yet, new connections from the client are still ending (sometimes) on node1 or node3 and not as often on node2 (monitoring this real time tonight).
I'm curious if the LOAD_BALANCE in the tnsnames file might be somehow overriding the Oracle service load balancing?
Thoughts?
Chris Taylor
Oracle DBA
Parallon IT&S
-- http://www.freelists.org/webpage/oracle-lReceived on Thu May 16 2013 - 09:44:10 CEST