ableasdale / kip-714

Testing KIP-714

Home Page:https://cwiki.apache.org/confluence/display/KAFKA/KIP-714%3A+Client+metrics+and+observability

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

KIP-714: Client metrics and observability

Testing KIP-714: Client metrics and observability

Prerequisites

This project requires the use of Apache Kafka 3.7.0 or later.

Getting Started

The provided docker-compose file will initialise a single Apache Kafka broker (running in KRaft mode).

The broker instance has been configured to use a pre-configured broker plugin for the metrics (io.confluent.cse.KafkaClientTelemetry).

This plugin needs to be provided to the Docker container before you can start the container up; if you start the container without creating the jar file containing the plugin, the broker will fail to start, logging a org.apache.kafka.common.KafkaException: Class io.confluent.cse.KafkaClientTelemetry cannot be found message.

To build the jar, run:

gradle shadowJar

To start the instance, run:

docker-compose up -d

Let's connect to the broker and look around at the new feature in more detail:

docker-compose exec broker bash

Let's start by looking at:

/opt/kafka/bin/kafka-broker-api-versions.sh --bootstrap-server broker:9092

You can review the Apache Kafka version running on the host instance (3.7.0) by running:

/opt/kafka/bin/kafka-broker-api-versions.sh --bootstrap-server broker:9092 --version

This runs:

/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server broker:9092 \
   --alter \
   --name 'basic_producer_metrics'\
   --metrics org.apache.kafka.producer. \
   --interval 5000

Confirm it's configured:

/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server broker:9092 --describe --name "basic_producer_metrics"

You should see:

Client metrics configs for basic_producer_metrics are:
  interval.ms=5000
  metrics=org.apache.kafka.producer.

Great - let's create a few batches of test messages to get started - exit from the terminal session to the broker docker instance if you're still connected and run:

gradle run

You should see telemetry metrics getting logged by the broker (every 5 seconds):

2024-05-22 21:47:17 [2024-05-22 20:47:17,487] [INFO] ***** exportMetrics *****
2024-05-22 21:47:17 [2024-05-22 20:47:17,487] [INFO] +++ CLIENT TELEMETRY: clientInstanceId=VQQTF4UpT8qqtooenFFlgA,

To access the logs:

docker logs broker

Accessing server.properties:

less /etc/kafka/docker/server.properties

Get root access to the container:

docker exec -u 0 -it broker bash

Less relevant stuff below...

TODO - add metrics - this works but it's the longer form version of the command:

/opt/kafka/bin/kafka-configs.sh --bootstrap-server broker:9092    --entity-type client-metrics  
  --entity-name "basic_producer_metrics"    --alter   
   --add-config "metrics=[org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max],
   interval.ms=15000,match=[client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538]"
/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server broker:9092 --list

Now let's describe it:

/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server broker:9092 --describe --name "basic_producer_metrics"

We should see:

Client metrics configs for basic_producer_metrics are:
  interval.ms=15000
  match=client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538
  metrics=org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max

Initial testing

docker-compose exec kafka bash

Or:

docker exec -ti kafka /bin/bash

****
./opt/kafka/bin/kafka-topics.sh
docker exec -ti kafka /opt/kafka/bin/kafka-topics.sh

Workings

docker-compose exec kafka /opt/kafka/bin/kafka-topics.sh --describe

docker-compose exec kafka /opt/kafka/bin/kafka-topics.sh --bootstrap-server kafka:9092 --describe

docker-compose exec kafka /opt/kafka/bin/kafka-topics.sh --bootstrap-server kafka:9092 --create

docker-compose exec kafka /opt/kafka/bin/kafka-topics.sh --bootstrap-server kafka:9092 --create --topic test-topic
Created topic test-topic.

docker-compose exec kafka /opt/kafka/bin/kafka-topics.sh --bootstrap-server kafka:9092 --list

Let's produce:

docker-compose exec kafka /opt/kafka/bin/kafka-console-producer.sh --bootstrap-server kafka:9092 --topic test-topic
kafka-configs.sh --bootstrap-server $BROKERS \
--entity-type client-metrics \
--entity-name "basic_producer_metrics" \
--alter \
--add-config "metrics=[org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max],interval.ms=15000,match=[client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538]"
kafka:/$ /opt/kafka/bin/kafka-configs.sh --bootstrap-server kafka:9092    --entity-type client-metrics    --entity-name "basic_producer_metrics"    --alter    --add-config "metrics=[org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max],interval.ms=15000,match=[client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538]"
/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server broker:9092 --list
/opt/kafka/bin/kafka-configs.sh --bootstrap-server kafka:9092 --describe --entity-type client-metrics --entity-name "basic_producer_metrics"
Dynamic configs for client-metric basic_producer_metrics are:
  interval.ms=15000 sensitive=false synonyms={}
  match=client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538 sensitive=false synonyms={}
  metrics=org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max sensitive=false synonyms={}

/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server kafka:9092 --describe --name "basic_producer_metrics"

Client metrics configs for basic_producer_metrics are:
  interval.ms=15000
  match=client_instance_id=b69cc35a-7a54-4790-aa69-cc2bd4ee4538
  metrics=org.apache.kafka.producer., org.apache.kafka.consumer.coordinator.rebalance.latency.max
docker-compose exec broker bash

/opt/kafka/bin/kafka-client-metrics.sh --bootstrap-server $BROKERS --describe

Notes

  • Ensure the Kafka Cluster is using KRaft mode
  • Write or obtain a metrics reporter that implements org.apache.kafka.server.telemetry.ClientTelemetry. Put that class on the class path for the brokers.
  • Change the broker configuration metric.reporters=<your class name>
  • Start the brokers.
  • Ensure that client telemetry APIs are supported using the kafka-broker-api-versions.sh tool
  • Create a client metrics resource to start collecting metrics using the kafka-client-metrics.sh tool
  • And then connect your clients (AK 3.7 or later)