Applying a cumulative fix to get the latest maintenance and in the case of 8.5 a set of new features is a common process and might be used every few months.
To reduce the downtime of the application the goal is to reduce the time it takes to apply the cumulative fix. Even a developer ideally wants the JVM to be unavailable for as a small time window as possible.
The following best practices and procedures can be leveraged to reduce the time it takes to apply a cumulative fix:
- Download the CF locally and leverage IIM with the downloaded CF
- Configure IIM to not contact the Internet to check for later software
- in a cluster after successfully applying the CF on the primary node run the application of the CF on the secondary nodes in parallel
- Speed up startup time of the Portal JVMs - apply Portal light mode or Portal developer mode in case it is a development system.
For details see: http://www-01.ibm.com/support/knowledgecenter/SSHRKX_8.5.0/mp/admin-system/portal_light_nbl.dita
- Perform a runstats/reorg on the database before applying the CF. In case you are running derby on a development system you can run the ConfigTask optimize-derby-database to trigger a runstats/reorg.
- Ensure enough (physical) CPU is assigned to the systems before applying the CF - including the Deployment Manager in case of a cluster - after the CF application the CPUs could be reduced again
- Ensure enough heap size is assigned to the Deployment Manager (recommending at least 2 GB) and nodeAgents (recommending at least 512 MB)
- Disable validation step in the CF: By default the first step in applying a CF is validating the credentials and other settings but it takes time to do this. To speed up the process the validation can be skipped by setting skip.cf.validation=true in wkplc.properties. Instead you can run the healtchcheck validation before applying the fix at another point in time.
The command would for instance be:
./ConfigEngine.sh action-health-check-validation -DRequiredWAS=188.8.131.52
For more details see either the readme of the CF you are trying to install or the following documentation (it refers to Portal 7 but still applies to 8.5): http://www-01.ibm.com/support/docview.wss?uid=swg27020200
- To reduce overall downtime (not related to how long it takes to apply a CF to a single JVM or cluster) consider a different deployment model - a farm or multi-cluster setup or high available cloud deployment (like Digital Experience on Cloud) can eliminate potential outages