Node Wise RAC Database / Cluster Update

A cluster update always requires that all the RAC database instances running on that node be shut down during the switch process of the cluster instance. A cluster update, followed by a RAC database update results in a database instance getting restarted twice.

There may be several situations where you as the administrator may require more control over the switch process. For example, you may choose to perform a node-wise update of the cluster and RAC databases in order to avoid multiple restarts of the database instances. You may also need to perform node specific pre and post steps.

This option provides this control by enabling you to perform the following tasks for each node:

• Switch the cluster instance
• Leave the RAC database instances shutdown

  For example, RAC databases RACDB_112 and RACDB_121 are running on this cluster. The instances RACDB_112_1 and RACDB_121_1 running on this specific node will continue to remain shut down after the cluster instance is switched.

  emcli db_software_maintenance 
  -performOperation 
  -name="Update Cluster" 
  -purpose=UPDATE_GI 
  -target_type=cluster 
  -target_list= CLUSTER1 
  -normal_credential="NC_HOST_CREDS:SYSMAN" 
  -privilege_credential="HOST_PRIV:SYSMAN" 
  -rolling=true -node_list="host1.example.com" 
  –startupDatabase=false
  -skip_cvu_check=true

• Switch all the RAC database instances on the same node.
• This step will switch the instances RACDB_112_1 and RACDB_121_1 to the new home and will restart the same.

  For example, consider RAC databases RACDB_112 and RACDB_121 are running on this cluster. The instances RACDB_112_1 and RACDB_121_1 running on this specific node will continue to remain shut down after the cluster instance is switched.

• SKIP_CVU_CHECK: This flag allows skipping the CVU check if set to true. This is an optional flag in UPDATE_GI operations.
• When performing UPDATE_GI operations and you need to pass multiple clusters in the -target_list, -node_list is not supported.