Migrating RAC Nodes [message #466522] |
Tue, 20 July 2010 00:47 |
kumarrajnishgupta
Messages: 43 Registered: October 2008 Location: noida
|
Member |
|
|
Hi,
I am having 3 nodes RAC cluster which is installed on Centos 4.5 with Oracle 10.2.0.4.0. Now due to some limitation will have to migrate this all 3 nodes with Rhel 5 ( latest version ). Could i add a node which installed with rhel5 + oracle 10.2.0.4.0 one by one for replacing/migrating on rhel5 without downtime. or As per you what would be the best approach for this. Also we have physical stand by for the same.
please advice us.
Regds
Rajnish
|
|
|
Re: Migrating RAC Nodes [message #466949 is a reply to message #466522] |
Wed, 21 July 2010 14:26 |
mkounalis
Messages: 147 Registered: October 2009 Location: Dallas, TX
|
Senior Member |
|
|
I would assume you are migrating to move to a supported OS. I hope you have a development system that mimics your production system. If so, I do think your plan of adding in the RHEL5 nodes and then removing the CENTOS nodes would be your best chance for mitigating downtime. I wouldn't advocate this though without testing your processes. Alternatively you could fail over to your physical standby, rebuild your cluster on RHEL 5, make the new RHEL 5 cluster your physical standby for your now current production system, then switch over to your RHEL 5 system. Whatever you decide to do, TEST FIRST.
|
|
|
|
Re: Migrating RAC Nodes [message #467217 is a reply to message #466990] |
Thu, 22 July 2010 12:46 |
mkounalis
Messages: 147 Registered: October 2009 Location: Dallas, TX
|
Senior Member |
|
|
This is where testing is important. If you are running the same exact bitness (i.e. 64-bit Linux OS) on both systems, and you are running the same Oracle binary installations on both systems, I think you are probably going to be OK to do this. Would having different OS's on your primary and standby in production be a good idea? No. But - I think you would be OK for the duration of your switchover. After going to your new RHEL 5 system then you would re-build the standby with RHEL and re-instantiate your standby database. Again - this shouldn't be attempted first in your production environment - test test test test test.
|
|
|