Thought I had copied to the group..
===
Hi Robert,
We are not using shared GI/OH binaries.
My Prod environment has 12 nodes, and we have multi-node cluster in HA and DR
environments as well. I am trying to eliminate manual binary install in each of
the nodes..
Would the following method work?
1. Go through software only install for GI/OH and put in 11.2.0.4 (as you had
specified)
2. Clone the new binaries as described in the doc: How to Clone an 11.2.0.3
Grid Infrastructure Home and Clusterware (Doc ID 1413846.1).
3. Copy the cloned binaries into the other nodes and get everything put in place
4. Go through the downtime to complete the rest of the steps on all the nodes
A follow up question would be:
Trying to see how my rollback plan would look like? To simplify it, if we run
into issues, my preference would be to troubleshoot it in prod unless database
down or application not functional scenarios.
I am looking for rollback scenario - before the application data is modified
and we run into any unknown issues during the upgrade..
Thanks again
-Upendra
From: rfreeman@xxxxxxxxxxxxxxxxx
To: nupendra@xxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
Subject: RE: Oracle GI patching
Date: Wed, 20 Apr 2016 21:27:55 +0000
Absolutely you can do the same thing with the Grid Infrastructure. The high
level steps are as follows:
1.
Lay down the 11.2.0.4 GI patch set (Number 3 of the 7 file patchset) using the
OUI. Select the install software only option. As I recall, when using
this option you need to install the software on each node of the cluster if
you are not using a shared OH – the OUI does not copy it across for you like it
would for an install/configure operation… So, you just repeat this process on
each node.
2.
Then, download and install the most current version of OPatch and install it in
your new 11.2.0.4 GH directory.
3.
Now, get the latest PSU, unzip it and use OPatch to apply it to the GH
directory. You will need to run opatch napply three times. One for the OCW
patch,
one for the ACFS patch and one for the DB patch. Something like this:
$GRID_HOME/OPatch/opatch napply -oh $GRID_HOME -local
$PATCH_LOCATION/$PATCH_BASE_NUMBER/$PATCH_OCW_NUMBER
4.
Then, when you are ready to actually upgrade the cluster, you will run
config.sh from the new GRID_HOME. It will upgrade each node of the cluster, one
node at a time – so a rolling upgrade.
There are a mass of documents that you can use to kind of piece all of this
together. Some of them include:
1.
The readme documents for the 11.2.0.4 base and the quarterly patch.
2.
How to Configure or Re-configure Grid
Infrastructure With config.sh/config.bat (Doc ID 1354258.1)
3.
Section F7 of the GI Install guide
4.
Oracle Grid Infrastructure 11.2.0.4.x Patch Set Update SUPPLEMENTAL README (Doc
ID 1641136.1) – This is a really important one I think – it addresses
various install situations, including the one you describe.
Hope that helps….
RF
Lay down the 11.2.0.4 base patch (it’s a full install) using the install
software only option of the OUI.
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
On Behalf Of Upendra nerilla
Sent: Wednesday, April 20, 2016 3:23 PM
To: Oracle-L <oracle-l@xxxxxxxxxxxxx>
Subject: Oracle GI patching
Hello everyone -
We are running OEL-6, Grid Infrastructure 11g, Oracle 11g. We are upgrading
from 11.2.0.3 to 11.2.0.4 on both GI and Database versions.
It is out of place upgrade. I am trying to minimize the total downtime during
the upgrade process.
From the database perspective, here is what we are doing to upgrade:
* Install the binaries ahead of the time
* At the time of the maintenance, switch to new Oracle home and complete
running catupgrade scripts
Is there a similar option for Grid Infrastructure? Could anyone point me to any
documents or share their notes..
Are there any pitfalls or risks?
Thanks much in advance
-Upendra
