Updating your Business Automation Insights custom resource

You can update your IBM Business Automation Insights custom resource (CR) by passing new configuration parameter values in the YAML file and running the oc apply command. You do not need to scale down operators, nor to delete Kubernetes jobs and pods.

About this task

Important:
  • Updates to bai_configuration can trigger Flink job restarts and may cause BAI data inconsistency. Before making configuration changes (such as enabling/disabling jobs, filters, emitters, or changing parallelism), ensure that either Flink High Availability (HA) is enabled or a savepoint-based recovery path is configured. For more information, see Restarting from a checkpoint or savepoint and Enabling Flink Job Manager high availability.
  • For some parameters, you cannot change the value that you set at initial installation. For a full list of the configuration parameters, see the reference tables in Business Performer Center parameters.

Updates to the custom resources (CR) overwrite the deployed resources during the operator control loop (observe, analyze, act), which occurs as a result of this loop constantly watching the state of the Kubernetes resources. If you choose to modify the installed software, use the same CR YAML file that you deployed with the operator to make the updates (for example, my_bai_cr.yaml).

In code lines, the NAMESPACE, bai_namespace, or ${NAMESPACE} placeholder is the namespace where Business Automation Insights is deployed. The CR_NAME or custom_resource_name placeholder is the name of the custom resource that was used to deploy IBM Business Automation Insights.

Procedure

  1. Edit your custom resource YAML file to include all your intended modifications.
  2. Apply the updates to the operator. 
    oc apply -f /my_bai_cr.yaml --overwrite=true
    Note: You can also edit YAML files by using the default UNIX visual editor (vi).
    oc edit InsightsEngine <custom_resource_name>
  3. Verify the updated automation containers.

    The operator reconciliation loop might take several minutes.

    1. Monitor the status of your pods. 
      oc get pods -w
    2. When all of the pods are in Running state, access the status of your services. 
      oc status