Here are the steps I use to deploy an Application to a Websphere Cluster.
- Get a list the Servers that belong to the Cluster
- Loop through the list of Servers in the Cluster
- Stop the App on each server
- Remove the App
- Install the App
- Apply warClassLoaderPolicy updates
- Apply classLoader updates
- Apply startingWeight
- Loop through the list of Servers in the Cluster and apply the following if changed
- Stop the Server
- Apply genericJvmArguments
- Apply initialHeapSize
- Apply maximumHeapSize
- Apply customProperty
- Apply websphereVariables
- Start the Server
- Loop through the list of Servers in the Cluster
- Start the Server if needed
- Start the App on the Server if needed
TIPS:
The stopping and re-starting of the Server is only needed if genericJvmArguments, initialHeapSize, maximumHeapSize, customProperty, websphereVariables values have changed. Otherwise, the servers can be left running saving total deployment time.
Make sure to check if something is stopped or started prior to stopping or starting it, this will prevent errors from being thrown.