7.6 Upgrading Existing Compute Node Configuration from Release 1.0.2

When you run a Software Update on an existing Oracle PCA installation with controller software Release 1.0.2, you must execute an additional upgrade procedure on provisioned compute nodes. The procedure is designed to reconfigure the compute node storage connections and the Oracle VM Agent, allowing the compute node to continue to function reliably as a member of the clustered server pool after the software update. By following the instructions in this section, you also make sure that previously deployed virtual machines remain fully functional when the appliance update to Release 2.0.1 or newer is complete.

Warning

First update any Release 1.0.2 system to appliance software Release 1.1.3. Execute the procedure in this section on each compute node provisioned with software Release 1.0.2, after the Release 1.1.3 software update on the management nodes has completed successfully. Proceed with the update to Release 2.0.1 or newer only when these steps have been completed.

Note

A large part of this procedure must be executed from Oracle VM Manager. An overview of its usage in the context of Oracle PCA, and links to detailed instructions in the Oracle VM documentation, are provided in this Administrator's Guide, in Chapter 4, Managing the Oracle VM Virtual Infrastructure.

Upgrading the Compute Node Configuration from Release 1.0.2 to Release 1.1.3

Make sure that the appliance software has been updated successfully to Release 1.1.3.
You can verify this by logging into the master management node and entering the following command:
```
# pca-diag version
1.1.3_<build>_<date>
```
Log in to Oracle VM Manager.
For details, see Section 4.2, “Logging in to the Oracle VM Manager Web UI”.
Identify the master server in the server pool.
1. In the Servers and VMs tab, select the server pool in the navigation pane.
2. In the management pane, select the Info perspective from the list in the toolbar.
  A configuration overview of the server pool is displayed.
3. Locate the Master Server setting and write down which server is the pool master.
  Warning
  The server pool master server must be the last Oracle VM Server to go through this upgrade procedure.
Migrate all running virtual machines away from the compute node you want to upgrade.
Information on migrating virtual machines is provided in the Oracle VM User's Guide section entitled Migrating Virtual Machines.
Place the compute node in maintenance mode.
Information on maintenance mode is provided in the Oracle VM User's Guide section entitled Placing an Oracle VM Server into Maintenance Mode.
1. In the Servers and VMs tab, select the Oracle VM Server in the navigation pane. Click Edit Server in the management pane toolbar.
  The Edit Server dialog box is displayed.
2. Select the Maintenance Mode check box to place the Oracle VM Server into maintenance mode. Click OK.
  The Oracle VM Server is in maintenance mode and ready for servicing.
Install the ovca-support package by executing the steps described in Section 7.4, “Enabling Support Tools on Systems Provisioned with Software Release 1.0.2”. However, do not log out of the compute node yet.
From the Oracle Linux command line on the compute node, run this script:
```
# /usr/sbin/cn_upg_111.sh
```
When the upgrade script completes successfully, log out of the compute node and return to the Oracle VM Manager UI.
Run the Oracle VM Server update on the compute node in question.
Information on the Oracle VM Server update functionality is provided in the Oracle VM User's Guide section entitled Updating and Upgrading Oracle VM Servers.
1. In the Servers and VMs tab, select the server pool in the navigation pane.
2. In the management pane, select the Servers perspective from the list in the toolbar.
  A table is displayed with all servers in the server pool.
3. Select the appropriate compute node in the management pane table and click Update Server in the management pane toolbar.
  As part of the update procedure, the Oracle VM Server is restarted but remains in maintenance mode.
Warning
If the compute node does not reboot during the update, you must restart it from within Oracle VM Manager.
Take the compute node out of maintenance mode.
1. In the Servers and VMs tab, select the Oracle VM Server in the navigation pane. Click Edit Server in the management pane toolbar.
  The Edit Server dialog box is displayed.
2. Clear the Maintenance Mode check box. Click OK.
  The Oracle VM Server rejoins the server pool as a fully functioning member.
If the compute node is in WARNING status, unpresent the Rack1-Repository from this node, and remove the node from the Rack1_ServerPool.
1. In the Repositories tab, select the Rack1-Repository in the management pane. Click Present/Unpresent Repository in the management pane toolbar.
  The Present Repository to Servers dialog box is displayed.
2. Select the Servers option at the top of the dialog box. Move the server in question out of the Present to Server pane. Click OK.
  The repository is unpresented from the selected Oracle VM Server.
3. In the Servers and VMs tab, select Rack1_ServerPool and click Edit Server Pool.
  The Edit Server Pool dialog box is displayed.
4. In the Servers tab of the Edit Server Pool dialog box, move the server in question out of the Selected Servers pane. Click OK.
  The Oracle VM Server is removed from the server pool.
Repeat this procedure for each compute node that requires upgrade from Release 1.0.2 to Release 1.1.3.
Warning
The server pool master server must be the last Oracle VM Server to go through this upgrade procedure.
When all compute nodes have been upgraded and rebooted successfully, acknowledge all events for the Oracle VM Servers and the Rack1-Repository.
Information on acknowledging events is provided in the Oracle VM User's Guide section entitled Acknowledging Events/Errors.
Remove the stale LUN mappings left behind from the previous storage configuration.
1. Go to the Storage tab of the Oracle VM Manager UI.
2. In the navigation pane, select the SAN Server named OVCA_ZFSSA_Rack1.
3. In the management pane, select the LUN or LUNs in WARNING status and click Delete Physical Disk in the toolbar.
Re-add any node that was removed from the Rack1_ServerPool. Make sure the Rack1-Repository is presented to all servers in the pool.

If no errors have occurred during this procedure, the software update to Release 1.1.3 is now complete.