In this blog, I’ll walk you through the exact step-by-step patching plan you can follow in a multi-node Exadata setup, covering everything starting by RoCE switch, Exadata Storage, KVM hosts, KVM Guest as well as GI & Oracle Homes patching.
In this patch plan I will show you a solid patching workflow that minimizes downtime and ensures smooth integration of all hardware and software layers. Enjoy..
Environment Overview
This patching plan is based on an Exadata environment consisting of:
- 4 KVM Hosts (
KVM1toKVM4)
- 12 KVM Guests (3 VMs in each KVM host)
- 7 Cell Servers
- 2 RoCE Switches
Patching Sequence
1. Patch GI & RDBMS Homes
Start by applying the latest updates to Grid Infrastructure and Oracle Database homes. Keeping the Oracle stack updated first ensures database stability before hardware and network updates.
2. Patch RoCE Switches
As the central access point, KVM1 initiates the RoCE switch firmware upgrade using patchmgr. This step is performed first to ensure reliable fabric connectivity before touching compute or storage nodes.
I would like to mention, in this step we can include the Patch of Cisco Management Switch now or delay it to a different time.
3. Patch All Cell Servers from KVM1
Using patchmgr, apply the cell image updates to all 7 storage servers from KVM1. This step includes a rolling update and verification checks to maintain availability.
4. Patch KVM1 & KVM2 from KVM3
We will use KVM_3 to perform the patch for KVM_1 and KVM_2.
5. Patch All VMs Guest in KVM1 & KVM2
As you will see, I have used VM_3 to patch VM_1 and VM_2 for each RAC environment. This includes OS patches,Grid Infrastructure and Oracle Database Homes
