Overview of Business Automation Insights architecture and basic installation checks in Cloud Pak for Business Automation

Steps

1. BAI Architecture diagram

BAI architectural flow illustrates how operational events from multiple CP4BA components such as Workflow, Case, Content, and Decisions are collected and routed into a unified analytics pipeline. Event emitters send raw business and technical events to Kafka, which provides durable and scalable event ingestion. Apache Flink then processes these events through component‑specific jobs (e.g., BPMN, Case, ADS) to generate time‑series analytics and process summaries. The processed insights are stored in OpenSearch, where indices and aliases are automatically created during installation. Finally, Business Performance Center (BPC) and external dashboards visualize this data, allowing business teams to monitor process performance and KPIs in real time.

• dba/bai-bpmn – Processes events from the BAW BPMN component
• dba/bai-icm – Processes events from the Case Management component
• dba/bai-navigator – Processes events from the Content Navigator component
• dba/bai-ads – Processes events from the Automation Decision Services component
• dba/bai-content – Processes events from the Content Management component

2. Steps to confirm that BAI Capability Is Deployed in CP4BA

2.1 Follow all tasks documented in IBM KC for Preparing to install Business Automation Insights. After preparation, Configure BAI for event processing, Ensure BPMN processing is enabled and Complete post-installation tasks.

2.2 Enable BAI BPMN Processing in the CP4BA Custom Resource YAML

bai_configuration:

  bpmn:

    install: true

    force_elasticsearch_timeseries: true

This confirms BAI to process BPMN events generated by BAW.

2.3 From IBM KC Configuring the BPM event emitter for BAW on containers

workflow_authoring_configuration:

  business_event:

    enable: true

    enable_task_api: true

    enable_task_record: true

    subscription:

      - app_name: '*'

        component_name: '*'

        component_type: '*'

        element_name: '*'

        element_type: '*'

        nature: '*'

        version: '*'

Key requirement is to have business_event.enable: true. BAI cannot receive workflow events unless the emitter is enabled.

2.4 Confirm BAI Capability Status in CR

   bai:

     bai_deploy_status: Ready

     insightsEngine: Ready

2.5 Confirm CP4BA capabilities include BAI

   bastudio:

     capabilities:

       - workflow

       - content-integration

       - case

       - bai

       - baml

2.6 After updating CR parameters, wait for operator reconciliation to complete.

3. Verifying BAI pods and endpoints to access apps.

3.1 Retrieve BAI and BAW component URLs from the Access ConfigMap icp4adeploy-cp4ba-access-info

Screenshots showing the initial data on BAI dashboard via https://cpd-cp4ba.ocp.com/bai-bpc

Follow this technote "Enabling and Validating BAW Business Data in BAI Dashboards on CP4BA" to get the business data visible on BAI dashboard. 

4. Troubleshooting: BAI 

This section summarizes the key troubleshooting areas relevant to BAW → BAI event flow. For detailed troubleshooting steps, refer to the official IBM documentation: Troubleshooting IBM Business Automation Insights on Kubernetes .

a) Troubleshooting the Workflow Event Emitter

If BAI dashboards do not show workflow execution or tracking data, the first component to verify is the BAW Event Emitter. IBM provides a dedicated troubleshooting guide:

Key checks include:

Ensure business_event.enable: true in the CP4BA Custom Resource (CR).

Enable tracing temporarily in the CR YAML to diagnose issues with event generation or publishing.

Review Workflow Runtime (BAW runtime) pod logs to ensure events are being emitted without errors.

This helps confirm whether the source (BAW) is generating events before checking downstream BAI components.

b) Troubleshooting Apache Flink Jobs

Flink is responsible for transforming raw workflow (BPMN) events into time‑series data and summaries for display in BAI dashboards. If Flink is not consuming events, BAI cannot display data. Official guidance from IBM.
 

Check Running Flink Jobs

Run the following commands to list all current Flink jobs and verify the status of the BAW BPMN job (dba/bai-bpmn):

JOB_MANAGER_POD=$(oc get pods -l component=jobmanager --no-headers -o custom-columns=Name:.metadata.name | head -n 1)

oc exec -c flink-main-container -it ${JOB_MANAGER_POD} -- \

/opt/flink/bin/flink list | grep RUNNING

If dba/bai-bpmn is not in RUNNING state, events will not reach OpenSearch.

Enable Verbose Raw Event Logging (Optional)

To diagnose event processing issues inside Flink itself, you can enable verbose logging:

     bai_configuration:

        bpmn:

          install: true

          force_elasticsearch_timeseries: true

        flink: 

          verbose_logs: true

This helps validate whether raw events are being consumed, processed, and forwarded correctly.

Refer to the documentation: Restarting Flink jobs after failure .
 

c) Accessing OpenSearch to Validate Indexed Data

BAI dashboards read data from OpenSearch. If data exists in Kafka and Flink completes processing but dashboards still show nothing, verify that documents are being indexed.

(i) Install Elasticvue browser extension, a free GUI for Elasticsearch/OpenSearch.

(ii) Obtain the OpenSearch URL and credentials from the CP4BA access ConfigMap and add the given open search cluster in Elasticvue. 

OpenSearch URL: https://opensearch-cp4ba.apps.cp4ba.ocp.com

Connection secret: opensearch-admin-user

(iii) Search for indexes such as below under Indices 

icp4ba-bai-process-*

icp4ba-bai-activity-*

icp4ba-bai-timeseries-*

Document count increasing when new BAW instances are executed.

Presence of your process application name (e.g., Hiring Sample) in indexed event documents.

If data is missing from OpenSearch, Flink or Kafka may not be processing events.

Example: