Author: Tony Wu

This document described how to leverage the OVF template to export the Oracle RAC VMs to accelerate the Oracle RAC deployment in a Virtual SAN environment and to reduce the efforts to reinstall the whole environment.

In a Virtual SAN environment, you need to enable Multi-Writer, however this option will not apply to the deployed VM by using OVF template. Therefore, you need the disk format conversion. Also the network configuration will change, this will cause Oracle Clusterware does not work, especially for css daemon.  This article covered these parts.

Applied to: Native Virtual SAN environment, shared disks with multi-writer enabled for Oracle RAC

Export the VM from vCenter via Web Client of vCenter

Preparation:

1. Shut down all RAC VMs
2. Remove all non-boot disks from the VM with adding existing disks.

   Because all RAC VM shared one set of disks, we need find out which VM has the original disks created. This VM should be the first one when building RAC and other VMs should just added the “Existing Disks” when forming multi-node RAC, so you need remove all disks from the other VMs. In the context of this document, we use ORA1 as the VM with the first VM with the original disks added.

Export the VMs to local drive:

1. Go to the Template-> export OVF template of the VM. Or if you use C# client, go File->Export -> Export OVF template

Import the VM to the vCenter via Web Client

Preparation:

1. Make sure the VSAN SPBM is right created to the VM before deploy OVF template

Import the VMs to the vCenter

• Choose the OVF to deploy one by one

• In the select storage wizard page, choose the policy for Oracle VM, the policy needs have the rule to set the space reservation to 100 percent.
  

• After all RAC VMs deployed, don’t power them on, but remove all disks from the first VM except the boot disk.  Then open SSH to connect to one of the hosts in the Virtual SAN cluster. And go to the folder of the first VM – ORA1. You may list all VMDK used for the VM. And then use the vmkfstools to convert the zeroed thick disk to eager zeroed disk for the VM. 

vmkfstools -i "/vmfs/volumes/vsan:52d6f68442074f98-afc9bc63092d9f2a/8a1aa057-9dbe-6b1a-e71d-ecf4bbe94918/ora1_1.vmdk"

"/vmfs/volumes/vsan:52d6f68442074f98-afc9bc63092d9f2a/8a1aa057-9dbe-6b1a-e71d-ecf4bbe94918/ora1_1_convert.vmdk"-d eagerzeroedthick

        Refer the link - https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1033570for more information.

• After all disks converted, add all converted disks to first VM, with multi-writer enabled, and make sure the Independent –Persistent disk option is set, and the desired SCSI controller is selected.  See the figure below for the detailed settings.
• Add disks by using the same settings for the other VM when adding the “Existing disks” for the other VMs one by one.

Boot up VM and reconfigure the network

1. After power on all nodes, you need check the NIC name and IP address to make sure it is the same name and IP as the ones before exporting the VM. If the ETH name and IP address change, you need set them back
2. In order to change a network device name, you can thus edit /etc/udev/rules.d/70-persistent-net.rules. If there is no such file, you can create one yourself. Keep in mind that the last line will overwrite previous settings, so you only need record the MAC addresses of your new deployed VM and modify the last lines.

    

   In /etc/udev/rules.d/70-persistent-net.rules, you can assign an arbitrary device name to a particular MAC address. Thus, first find out the MAC address of your network interface, and then define the interface name to that MAC address in the following format.

    

   $sudo vi /etc/udev/rules.d/70-persistent-net.rules

   # interface with MAC address "00:0c:29:43:28:11" will be assigned "eth0"

   SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="00:0c:29:43:28:11", ATTR{dev_id}=="0x0", ATTR{type}=="1", KERNEL=="eth*", NAME="eth0"

    

   # interface with MAC address "00:0c:29:43:28:1b" will be assigned "eth1"

   SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="00:0c:29:43:28:1b", ATTR{dev_id}=="0x0", ATTR{type}

    

3. After reconfigurations, reboot the vm and check if Oracle RAC works normally.