Cluster maintenance

Maintaining IBM WebSphere Portal in a cluster means applying corrective services (fix packs and interim fixes) or 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 we apply any maintenance, analyze any user impact. Ensure we are able to provide uninterrupted service (also referred to as 24x7 availability), even during the maintenance phase.

Fixes are classified as "minor" if they do not update the underlying WebSphere Portal databases or require version upgrades to other supporting software such as databases servers or IBM WebSphere Application Server. Most of the WebSphere Portal service packs are not considered minor and might require the use of a separate installation procedure to ensure 24x7 availability.

If we are not using horizontal scaling in the environment, any fix that requires a restart results in temporary outage for the users. Existing 24x7 installation procedures do not apply to these environments.

Warning: We do not need to remove the node from the cluster when we apply fixes; doing so might result in the inability to add the node back to the cluster.

Minor fixes

Apply all minor fixes on each cluster node. Use the installation instructions that are supplied with the fix. When we apply minor fixes that might update previously deployed enterprise applications, first turn off the auto-synchronization feature of the deployment manager. After the fix is deployed on all cluster nodes, we can force a manual synchronization with the deployment manager to ensure that all updates are synchronized on the nodes. We can then enable the auto-synchronization feature again.

If the documentation associated with the minor fix requires that you restart WebSphere Portal or WebSphere Application Server, apply the minor fix one node at a time. This process enables other nodes to continue to provide service to the users. However, if the fix requires an update to the WebSphere Portal databases, we might be required to stop the cluster before we apply the fix. If so, use a procedure that ensures 24x7 availability.

Fixes and service packs

There are multiple approaches to installing service packs into a WebSphere Portal clustered environment. Installing a service pack involves modifying the IP sprayer to remove a cluster from receiving user requests. This process allows time to finish handling existing user sessions and to upgrading that cluster on all nodes. After verification testing assures the upgraded cluster is operational, it can start receiving production traffic while the next cluster is taken offline and goes through the upgrade process. This approach preserves complete 24x7 availability during the upgrade process.

A separate document is available describing the process of installing WebSphere Portal service packs (fix packs) into an existing cluster while you maintain 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. We then upgrade the node with the service packs. After the upgrade is complete, return the node or set of nodes to the flow of user traffic. Repeat the procedure with the next node or set of nodes. This process continues until all nodes in the cluster are upgraded.

While the upgrade process is taking place, some portlets might become temporarily unavailable because of updates to the shared database, which are incompatible with the previous version of the portlet. This process can introduce functional limitations to the 24x7 availability.

Parent Cluster considerations