Documentation Grafana Cloud Monitor applications Asserts Enable metrics collection Messaging frameworks Kafka

Configure Kafka exporter to generate Prometheus metrics

You can configure Kafka server to generate Prometheus metrics either using an external exporter or JMX exporter.

To configure Kafka exporter to generate Prometheus metrics, complete the following steps:

1. Select one of the following methods:

   • To set up the Kafka exporter, refer to Exporter.

   • To set up JMX Exporter, refer to JMX Exporter.

     • If Kafka is running, you can use the following script as an alternative way to configuring the exporter:

       KAFKA_OPTS="$KAFKA_OPTS -javaagent:./jmx_prometheus_javaagent-0.16.1.jar=8080:./kafka-2_0_0.yml" kafka-server-start /usr/local/etc/kafka/server.properties

2. To confirm you configured Kafka exporter correctly, ensure the following metrics are available in Prometheus.

   • kafka_topic_partitions{topic="__consumer_offsets"}
   • kafka_topic_partition_current_offset gauge kafka_topic_partition_current_offset{partition="0",topic="__consumer_offsets"}

3. To confirm you configured JMX exporter correctly, ensure the following metrics are available in Prometheus.

   • kafka_producer_topic_record_send_total
   • kafka_producer_record_send_total
   • kafka_consumer_records_consumed_total_records_total
   • kafka_consumer_fetch_manager_bytes_consumed_total

RED metric KPIs

This section lists RED metrics KPIs.

Request rate

Asserts automatically tracks the following KPIs for your RED metrics.

• Kafka JMX RED metrics KPIs
  • Producer requests rate(kafka_server_brokertopicmetrics_totalproducerequests_total[5m])
  • Producer records rate(kafka_server_brokertopicmetrics_messagesin_total[5m])
  • Consumer requests rate(kafka_server_brokertopicmetrics_totalfetchrequestspersec_count{topic!=""}[5m])
• Kafka Exporter RED Metrics KPI
  • Produced messages avg_over_time((delta(kafka_topic_partition_current_offset{topic!=""}[1m]) > 0 or delta(kafka_topic_partition_current_offset{topic!=""}[1m]) * 0) / 60 [5m])
  • Consumed messages avg_over_time((delta(kafka_consumergroup_current_offset{topic!=""}[1m]) > 0 or delta(kafka_consumergroup_current_offset{topic!=""}[1m]) * 0) / 60 [5m])

Error ratios

• Producer errors rate(kafka_server_brokertopicmetrics_failedproducerequests_total{topic!=""}[5m])/rate(kafka_server_brokertopicmetrics_totalproducerequests_total[5m])
• Consumer errors rate(kafka_server_brokertopicmetrics_total_failedfetchrequestspersec_count{topic!=""}[5m])/ rate(kafka_server_brokertopicmetrics_totalfetchrequestspersec_count{topic!=""}[5m])

Latency

• P99 - Consumer Request kafka_network_requestmetrics_totaltimems{request="Fetch", quantile="0.99"} / 1000
• P99 - Consumer Group kafka_network_requestmetrics_totaltimems{request=~".*Group", quantile="0.99"}) / 1000
• P99 - Producer Request kafka_network_requestmetrics_totaltimems{request="Produce",quantile="0.99"} / 1000
• P99 - Broker Request kafka_controller_controllerchannelmanager_requestrateandqueuetimems{quantile="0.99"} /1000

RED metrics alerts

Asserts automatically tracks the short-term and long-term trends for request and latency for anomaly detection. Similarly, you can set thresholds for latency averages and P99 to record breaches. Asserts tracks error ratios against availability goals (default, 99.9%) and breaches (default, 10%).

Failure Alerts

• KafkaTopicsUnderReplicatedPartitions kafka_topic_partition_under_replicated_partition > 0
• KafkaOfflinePartitions kafka_controller_kafkacontroller_offlinepartitionscount > 0
• KafkaActiveController kafka_controller_kafkacontroller_activecontrollercount != 1
• KafkaUnderMinIsrPartitions kafka_cluster_partition_underminisr > 0

Dashboards

The following dashboard shows information about Kafka server metrics:

• Messages Produced
• Messages Consumes
• Lag by Consumer
• Partitions for Topics

RED metrics - Producer

This section lists Producer RED metrics.

Requests

• Producer Record rate(kafka_producer_record_send_total[5m])
• Producer Requests rate(kafka_producer_request_total[5m])

Error ratio

• Producer Record rate(kafka_producer_record_error_total[5m])`` /rate(kafka_producer_record_send_total[5m])

Latency

• Average max without(asserts_request_context)(kafka_producer_request_latency_avg/1000)

RED metrics - Consumer

This section lists Producer RED metrics.

Requests

• Consumer Record rate(kafka_consumer_records_consumed_total_records_total[5m])
• Consumer Requests rate(kafka_consumer_fetch_total_requests_total[5m])
• Consumer Fetch Requests rate(kafka_consumer_fetch_manager_fetch_total[5m])
• Consumer Fetch Record rate(kafka_consumer_fetch_manager_records_consumed_total[5m])

Latency

• Average max without(asserts_request_context) (kafka_producer_request_latency_avg/1000)

Alerts

Dashboards

The following dashboard captures information about both producer and consumer of Kafka client.

• Topics connected to producer/consumer
• Producer records
• Producer requests
• Producer latency
• Consumer records
• Consumer Lag