Collecting Kafka diagnostic data using Cloudera Manager actions

Learn how to collect Kafka diagnostic data using Kafka actions in Cloudera Manager. These actions provide quick access to diagnostic data for troubleshooting and support.

Use the Collect Kafka Cluster Diagnostics and Describe Kafka Topics actions in Cloudera Manager to collect Kafka diagnostic data. These actions are available in the Actions dropdown on the Kafka service and Kafka Broker role instance pages. Running these Kafka-specific actions is a more lightweight process compared to full diagnostic bundle collection, allowing you to quickly gather targeted data for analysis.

The actions do the following:

  • Collect Kafka Cluster Diagnostics – Collects detailed diagnostic information about the entire Kafka cluster, including topics, configurations, consumer groups, log directories, and offsets. The action uses the following command-line tools to gather data:

      • kafka-topics
      • kafka-configs
      • kafka-consumer-groups
      • kafka-log-dirs
      • kafka-get-offsets
      • kafka-transactions
      • kafka-broker-api-versions

    By default, data collected by the Collect Kafka Cluster Diagnostics action is included in the cluster-wide diagnostic bundles that Cloudera Manager collects automatically as well as the diagnostic bundles you capture manually with the Support > Send Diagnostic Data modal.

  • Describe Kafka Topics – Collects detailed information about all Kafka topics in the cluster. This action runs the kafka-topics command-line tool with the --describe option to gather data.

    This action provides more fine-grained data for detailed analysis of individual topics compared to cluster-wide diagnostic bundles.

Diagnostic data is printed to stdout, making it immediately available in the command modal after an action finishes. In addition, data is saved as a compressed archive on the cluster host where the action runs. The host where the action runs depends on which page you start the action.
  • Kafka service pageCloudera Manager chooses one of the Kafka Broker hosts to run the action. The name of the host is visible in the command modal. Start actions from this page if the host where data is saved is not important.

  • Kafka Broker role instance page – The action runs on the host of the Kafka Broker role instance. Start actions from this page if you want to save diagnostic data on a specific host.

The archive is saved to a subdirectory under /var/run/cloudera-scm-agent/process. The exact location is printed to stdout. Look for a log entry similar to the following to find the exact location of the archive.
Successfully created archive: /var/run/cloudera-scm-agent/process/1546339774-kafka-KAFKA_BROKER-CollectKafkaClusterInfoCmdBroker/info_collection/kafka_info_collection_2025_11_05__22_11_25.tar.gz