Storage Foundation for Oracle® RAC 7.4.1 Configuration and Upgrade Guide - Solaris

To prepare to upgrade SF Oracle RAC

Perform the steps on the first subcluster.

1. Log in as superuser to one of the nodes in the subcluster.
2. Back up the following configuration files on your system: main.cf, types.cf, CVMTypes.cf, CFSTypes.cf, OracleTypes.cf, OracleASMTypes.cf, , PrivNIC.cf, MultiPrivNIC.cf, CRSResource.cf, /etc/llttab, /etc/llthosts, /etc/gabtab, /etc/vxfentab, /etc/vxfendg, /etc/vxfenmode

   For example:

   # cp /etc/VRTSvcs/conf/config/main.cf \
   /etc/VRTSvcs/conf/config/main.cf.save

3. Installer verifies that recent backups of configuration files in VxVM private region have been saved in /etc/vx/cbr/bk.

   If not, a warning message is displayed.

   Warning:

   Back up the /etc/vx/cbr/bk directory.

4. Stop the applications that use VxFS or VxVM disk groups on each node, whether local or CFS.

   If the applications are under VCS control:

   # hagrp -offline grp_name -sys node_name

   If the applications are not under VCS control:

   Use native application commands to stop the application.

5. Switch over all failover service groups to the nodes in the other subcluster:

   # hagrp -switch grp_name -to sys_name

6. Stop the Oracle RAC resources on each node.

   • If the database instances are managed by VCS, take the corresponding VCS service groups offline. As superuser, enter:

     # hagrp -offline grp_name -sys node_name

   • If the database instances are not managed by VCS, then run the following on one node:

     For Oracle RAC 12c or later:

     $ srvctl stop instance -db db_name \
     -node node_name

     For Oracle RAC 11.2.0.2 and later versions of 11g Release 2:

     $ srvctl stop instance -d db_name \
     -n node_name

     For Oracle RAC 11.2.0.1 and earlier versions:

     $ srvctl stop instance -d db_name \
     -i instance_name

7. • If the Oracle database is managed by VCS, set the AutoStart value to 0 to prevent the database service group from starting automatically when VCS starts. Failing to perform this step results in the database attempting to come online after the upgrade; the attempt fails due to the presence of old libraries on the system.

     # haconf -makerw
     # hagrp -modify oracle_group AutoStart 0
     # haconf -dump -makero

   • If the Oracle database is not managed by VCS, change the management policy for the database to manual:

     For Oracle RAC 12c or later:

     $ srvctl modify database -db db_name -policy manual

     For Oracle RAC 11g:

     $ srvctl modify database -d db_name -y manual

8. Unmount all the CFS file system which is not under VCS control.

   # mount -v |grep vxfs | grep cluster

   # fuser -c /mount_point

   # umount /mount_point

   Make sure that no processes are running which make use of mounted shared file system or shared volumes.

   # fuser -cu /mount_point

9. Perform the following if the subcluster has non-global zones.

   • Make sure that non global zones are not running on the sub-cluster while performing Phase I Upgrade.

   • Update the zone configuration. For instructions, see the topic "Preparing to install non-global zones" in the document Storage Foundation and High Availability Solutions Virtualization Guide

10. Take all the parallel VCS service groups offline on each of the nodes in the current subcluster:

    # hagrp -offline grp_name -sys sys_name

11. If you plan to continue using the Storage Foundation for Databases (SFDB) tools, you must prepare to migrate the SFDB repository database before upgrading to SF Oracle RAC 7.4.1.