Kafka Integration

This integration guide describes how to monitor availability and performance of Apache Kafka with ATSD.

Step 1: Configure Collector

  1. Log in to the Collector web interface at https://collector_hostname:9443
  2. From the Jobs page select Import by expanding the split-button below the table.
  3. Import the kafka-jmx job.
  4. On the JMX Job page, enable the job status with the Enabled checkbox.
  5. Adjust the cron expression if required. For more information on cron expressions, see Scheduling.
  6. Select a target ATSD database to store data.
  7. Click Save.


Configure series collection

  1. Select kafka-series configuration.

  2. On the JMX Configuration page, enter the JMX connection parameters or use an Item List with predefined kafka parameters:

    • Host: Kafka hostname.
    • Port: JMX port.
    • User Name: JMX user name.
    • Password: Password for JMX user.
    • Entity: Optionally, specify the output of the hostname command on the Kafka server if it is different from kafka_hostname (for example if kafka_hostname represents a fully qualified name).
    • Other parameters are optional. For more information on JMX configuration, see JMX Job Documentation.
  3. Click Test to validate the configuration.

  4. Click Save.

Configure properties collection

  1. Select kafka-properties configuration.

  2. Set Host, Port, User Name, Password, and Entity fields as described in the previous section.

  3. Click Test to validate the configuration.

  4. Click Save.

Step 2: Configure Kafka in ATSD

  1. Log in to the target ATSD instance at https://atsd_hostname:8443.
  2. Navigate to the Metrics page and verify that jmx.kafka.* metrics are available.
  3. Navigate to the Entities page and verify that jmx.kafka.* properties are available for entities from kafka-properties configuration.
  4. Open the Settings > Entity Groups page and import Kafka entity group.
  5. Expand the Portals menu, click Configure and import Kafka portals (Enable Auto-enable New Portals).
  6. Open the Alerts menu, click Rules and import Kafka rules (Enable Auto-enable New Rules).
  7. Open the Entity Views menu, select Configure and import Kafka Entity View configuration.

Step 3: Verification

  1. Select an open Kafka on the Entity Views menu:

  2. Verify that the portal in the table header is available and refers to the Kafka portal:

  3. Verify that the portal for each entity refers to the Broker portal:

Consumer Lag

Consumer lag calculation requires information about producer offset and consumer offset.

Producer offset is collected from Kafka brokers by the JMX Job above.

Consumer offset is collected using a Kafka console consumer reading events from the __consumer_offset topic on one of the Kafka servers in the cluster.

Log in to the Kafka server.

Download the shell script to the Kafka bin directory.

# assign execute permission
chmod +x /opt/kafka_2.12-1.0.0/bin/send_offset.sh

For Kafka versions before use --zookeeper option instead bootstrap-server in the script.

Replace ATSD_HOST and TCP_PORT with actual values and launch the script.

The default ATSD TCP command port is 8081.

The script reads and sends topic offsets to ATSD under the hostname entity.

Launch the script:

nohup /opt/kafka_2.12-1.0.0/bin/send_offset.sh ATSD_HOST TCP_PORT &

If the hostname is different from the entity name used in the JMX job, specify the entity manually.

nohup /opt/kafka_2.12-1.0.0/bin/send_offset.sh ATSD_HOST TCP_PORT ENTITY &

The script continuously reads consumer offsets from Kafka and sends the offsets to ATSD as series commands. Kafka also copies the commands stdout for debugging.

series e:nurswgvml702 m:kafka.consumer_offset=455 t:groupid="console-consumer-72620" t:topic="test" t:partition=0 ms:1519893731570
series e:nurswgvml702 m:kafka.consumer_offset=492 t:groupid="console-consumer-72620" t:topic="test" t:partition=0 ms:1519893736569
series e:nurswgvml702 m:kafka.consumer_offset=492 t:groupid="console-consumer-72620" t:topic="test" t:partition=0 ms:1519893741570
series e:nurswgvml702 m:kafka.consumer_offset=550 t:groupid="console-consumer-72620" t:topic="test" t:partition=0 ms:1519893746570
  1. Check that metric kafka.consumer_offset is available on the Metrics tab in ATSD.
  2. Import Consumer Lag Portal into ATSD and change the topic name to view the consumer lag.

For additional Kafka integration, see Brokers Monitoring and Consumers Monitoring.