Cluster maintenance
Maintaining IBM WebSphere Portal in a cluster typically means...
- applying corrective services (fix packs and interim fixes)
- updating the software release level on each node in the cluster
Instructions for applying corrective service to a WebSphere Portal cluster are provided with the corrective service package. Before applying any maintenance, analyze any impact to end users and ensure that you are able to provide uninterrupted service (also referred to as 24x7 availability), even during the maintenance phase.
We classify fixes as "minor" if they...
- do not update the underlying WebSphere Portal databases
- do not require version upgrades to other supporting software such as databases servers or IBM WAS
Most of the WebSphere Portal service packs are not considered minor and may require the use of a separate installation procedure to ensure 24x7 availability.
If you have not implemented horizontal scaling in your environment, i.e. you have only vertical nodes in your cluster, any fix that requires a re-start will result in temporary outage for your end users. Existing 24x7 install procedures do not apply to these environments.
Minor fixes
All minor fixes to WebSphere Portal in a clustered environment can be deployed by simply applying the fix on each cluster node using the install instructions supplied with the fix. You do not need to remove the node from cluster to apply minor fixes. When applying minor fixes that might update previously deployed enterprise applications, be sure to turn off the auto-synchronization feature of the deployment manager before applying the fix. After the fix is deployed on all cluster nodes, you can force a manual synchronization using the deployment manager to ensure that all updates are synchronized on the nodes. You can then enable the auto-synchronization feature again.
If the documentation associated with the minor fix requires that WebSphere Portal or WAS be restarted, be sure to apply the minor fix one node at a time. This will enable other nodes to continue to provide service to your end users. However, if the fix requires an update to the WebSphere Portal databases you might be required to stop the cluster before applying the fix. If this is the case, use a procedure that ensures 24x7 availability.
Service packs
A separate procedure is available to install WebSphere Portal service packs (fix packs) into an existing cluster while maintaining 24x7 availability.
The procedure removes one site from the cluster while the other site takes the full user load. The removed site can then be upgraded with the fix packs. When the upgrade process is complete, traffic is routed to the upgraded site while the fix packs are installed on the other site. The two sites are then merged back into the original cluster.
A separate document is available which describes the process of installing WebSphere Portal service packs (fix packs) into an existing cluster while maintaining 24x7 availability. To briefly summarize this procedure, you remove a node or set of nodes from the flow of user traffic by configuring the IP sprayer and Web server. You then upgrade the node with the service packs. After upgrading is complete, you return the node or set of nodes to the flow of user traffic, while repeating the procedure with the next node or set of nodes. This process continues until you have upgraded all nodes in the cluster.
Parent topic
Cluster guidelines and limitations