In a dual-partition upgrade, you divide the cluster into two groups of nodes. You bring down one group of nodes and upgrade those nodes. The other group of nodes continues to provide services. After you complete upgrade of the first group of nodes, you switch services to those upgraded nodes. You then upgrade the remaining nodes and boot them back into the rest of the cluster.
The cluster outage time is limited to the amount of time that is needed for the cluster to switch over services to the upgraded partition, with one exception. If you upgrade from the Sun Cluster 3.1 8/05 release and you intend to configure zone clusters (Solaris 10 only), you must temporarily take the upgraded first partition out of cluster mode to set new private-network settings that were introduced in the Sun Cluster 3.2 release.
Observe the following additional restrictions and requirements for the dual–partition upgrade method:
ZFS root file system requirement – If you also upgrade the Solaris OS, do not use this upgrade method if the cluster uses ZFS for the root file system. Instead, you must use the live-upgrade method to upgrade the cluster.
Sun Cluster HA for Sun JavaTM System Application Server EE (HADB) - If you are running the Sun Cluster HA for Sun Java System Application Server EE (HADB) data service with Sun Java System Application Server EE (HADB) software as of version 4.4, you must shut down the database before you begin the dual-partition upgrade. The HADB database does not tolerate the loss of membership that would occur when a partition of nodes is shut down for upgrade. This requirement does not apply to versions before version 4.4.
Data format changes - Do not use the dual-partition upgrade method if you intend to upgrade an application that requires that you change its data format during the application upgrade. The dual-partition upgrade method is not compatible with the extended downtime that is needed to perform data transformation.
Location of application software - Applications must be installed on nonshared storage. Shared storage is not accessible to a partition that is in noncluster mode. Therefore, it is not possible to upgrade application software that is located on shared storage.
Division of storage - Each shared storage device must be connected to a node in each group.
Single-node clusters - Dual-partition upgrade is not available to upgrade a single-node cluster. Use the standard upgrade or live upgrade method instead.
Configuration changes - Do not make cluster configuration changes that are not documented in the upgrade procedures. Such changes might not be propagated to the final cluster configuration. Also, validation attempts of such changes would fail because not all nodes are reachable during a dual-partition upgrade.