Note the following when you apply batch corrections to the cluster system.
Back up the system environment before you attempt to apply a correction.
The software version to be installed on each node must be the same on all nodes in the cluster system. Also, the corrections must be the same on all the nodes constituting the system. Note, however, that this is not always true when rolling update, described below, is allowed.
To apply a batch correction, you must stop the node temporarily. This means that the job must be stopped, albeit temporarily. You should consider a maintenance plan to ensure that the maintenance is completed within a specified period. You must also examine the time and duration of the maintenance to minimize the impact on a job.
Rolling update is a method by which software is updated while the job continues to operate by executing job failover for a node in a cluster to separate the standby node from the operation in order to apply corrections to the node one by one.
If you apply this method, the job stop time required for software update can be minimized. To perform update with this method, however, you must satisfy the prerequisites for rolling update (the items to be corrected must be correctable with rolling update).
To apply this method, you must confirm the contents of the README file for the relevant patch and then contact field engineers.
To apply SRU, note the following points:
After the application of SRU, do not change the PRIMECLUSTER settings and configuration until the operating system is restarted.
If SRU was applied at RMS startup, the cluster application may become Faulted. In this case, follow the procedure in "7.2.2.4 Bringing Faulted Cluster Application to Available State" to clear the Faulted cluster application to Online.
