Storage Foundation for Oracle® RAC 8.0.2 Configuration and Upgrade Guide - Linux

To perform a rolling upgrade in a CVR environment

1. Perform a rolling upgrade on the secondary sites first, without stopping the replication.

   Check whether the cluster nodes have joined the cluster back after the upgrade.

   Continue to monitor the replication status.

   # vradmin -g <disk_group_name> -l repstatus <RVG_name>

2. Perform a rolling upgrade for all the nodes on the primary site by following these steps:

   At the primary site, the VxFS file systems are always mounted. Therefore, if you directly begin the upgrade, the installer returns the following error:

   CPI ERROR V-9-40-1480 Some VxFS file systems are mounted 
   on mount points /<volume_mount_point> on node <node_name> 
   and need to be unmounted before upgrade.

   In this event, take the applications offline and unmount the file system manually on the indicated nodes, before you upgrade them.

   Additionally, you may have to take the following actions on these nodes:

   • If any parallel service groups exist in the hierarchy, take them offline.

     # /opt/VRTS/bin/hagrp -offline <global_application_name> -sys <node_name>

     # /opt/VRTS/bin/cfsumount /<volume_mount_point> <node_name>

     Make sure that CVM is the final child in the hierarchy, that it is left online, and rest of the hierarchy is taken offline on that node.

   • If any parent service groups exist in the hierarchy, take them offline.

     # /opt/VRTS/bin/cfsumount /<volume_mount_point> <node_name>

     # /opt/VRTS/bin/hagrp -dep cvm

     Parent Child Relationship
     <global_application_name> cvm online local firm

     <global_application_name> is the parent service group of the CVM; take it offline.

     # /opt/VRTS/bin/hagrp -offline <global_application_name> -sys <node_name>

   • If any local service groups exist in the hierarchy, switch them to another cluster node that is not being upgraded at the same time.

   • After you upgrade these nodes, mount the file system again and bring the VCS application online on the targeted node.

     # /opt/VRTS/bin/cfsmount /<volume_mount_point> <node_name>

     # /opt/VRTS/bin/hagrp -online <global_application_name> -sys <node_name>

   • Consequently, you may have to take the following actions on these upgraded nodes:

     • Bring the parallel service groups online.

     • Switch the local VCS service groups back.

     • Make sure to bring all the service groups, which were taken offline before the upgrade, online again.

     • Check the replication status and the system or the service group state, and only then proceed to perform the upgrade on the next node.

3. Check the replication status on the primary site.
4. Migrate primary role to the upgraded secondary and then proceed with upgrade on the secondary (old primary).

To upgrade disk group and disk layout versions on replication hosts

1. Upgrade the disk group version on all the Secondaries for all the disk groups.

   # /usr/sbin/vxdg upgrade <disk_group_name>

2. Upgrade the disk group version on the Primary for all the disk groups.

   # /usr/sbin/vxdg upgrade <disk_group_name>

3. Upgrade the disk layout version (DLV) on the Primary for all the VxFS file systems.

   # /opt/VRTS/bin/vxupgrade -n 17 <vxfs_mount_point_name>

   # /opt/VRTS/bin/fstyp -v <disk_path_for_mount_point_volume>

   The DLV upgrade is automatically replicated to the Secondaries.