This section describes Spark-related procedures that are required for configuration changes in KPI Management.
Updating the Spark Cluster- or Application Configuration
- Stop the KPI Management workflows
- Kill the Spark application
- Stop the Spark cluster (mzsh spark cluster-shutdown)
- Update the Spark service configuration
- Apply the changes (mzsh spark cluster-update)
- Make whatever changes are needed
- Start the Spark cluster
- Submit the Spark application
- Start the KPI Management workflows
| Note | ||
|---|---|---|
| ||
When you have changed properties in the configuration related to the service model or Kafka, you also need to clear the check-point directory before starting the Spark cluster. |
Updating the Kafka or Zookeeper Configuration
- Stop KPI Management workflows
- Kill the Spark application
- Shutdown the Platform and SCsSpark, Kafka and Zookeeper
- Clear all the runtime data.
- Update the Kafka and/or Zookeeper service configuration
- Start the PlatformZookeeper, SCs Kafka and servicesSpark
- Submit the Spark application
- Start the KPI Management workflows
...