I must thank my fellow DBA Franky Weber Faust for his publication in Portuguese OTN.

Apply the PSU in Standby with RAC database using the standby-first patch apply method.

In this article, I introduce how to apply the patch using the standby-first patch apply the method in a DataGuard environment with Oracle RAC 12c, where we apply with Rolling Patch. In short, we will first apply the patch to standby and because the database is a 2-node RAC, we apply it to one node first and then the other.

Scenario:

Primary:
Exadata Database Machine:
Srv-pmy1 → pmy1
Srv-pmy2 → pmy2
Standby:
Exadata Database Machine:
Srv-stb1 → stb1
Srv-stb2 → stb2

First of all, read the README of the patch that will be applied and make sure the patch can be applied the way I am doing here.

Initially, check the patches applied.

In the scenario that I am, there is a patch 21923026 incompatibility with the patch 24448103 that I will apply, so it is necessary to roll back the patch 21923026 before applying the other.

Check the running instances.

Connect to the Broker and check the current Standby configuration.

Thus it was verified that redoapply is occurring in node 2, which allows us to apply the patch in node 1. 
Stop all Oracle Home services where the patch will be applied. In my case, there is only one instance in this OH.

Do the patch rollback 21923026. The -verbose option brings more logs to the screen.

With the root user set the environment variables for Grid Infrastructure Home.

Run the opatchauto analyze to check the prerequisites.

Re-apply patch 21923026 that we had rolled back.

Before applying first check to see if there is any conflict.

Since we are not going to use the opatchauto for this patch, apply the patch in question only to the local node and when we finish this node do the same procedure on the other node.
Apply the patch with the -local option.

Now check the applied patches again. Note that the patch we apply is a Bundle Patch that is composed of 4 patches. The patch applied was BP 12.1.0.2.161018 which has the number 24340679.

Now go up the instance again.

Log in to the Broker and change the instance that is running redoapply.

Done that is just to repeat the whole procedure in node 2. I will not present here to not be repeated, but the step by step is the same.

References:

https://docs.oracle.com/database/121/HABPT/schedule_outage.htm#HABPT5321

Leave a Reply