Kafka Dlq Monitor

This small tool can be used to monitor a dead letter queue in the form of a kafka topic. It rings the alarm bells by logging an error to stdout. Depending on your logging stack (ELK or similar) this is enough to see a flag in your centralized logging system.

The following configuration snippets indicates how this application works.

topic: "in"
monitor: {
    duration: 1m
    max-dead-letters-in-duration: 500
}

It subscribes to a single topic and logs a warning if a certain threshold is reached within a given time period (e.g. 500 messages within 1 minute).

Everything can be configured easily with Hocon. Either via .conf file or as CONFIG_FORCE_* variables when running a container. The minimal configuration needed is presented in ./src/main/resources/application.conf

Local Kafka & Zookeeper

run

docker-compose up -d

Docker / Jib

Build the image for your local image repo by running

./gradlew jibDockerBuild

run container

docker run --network=host kafka-dlq-monitor

consider overriding configs when running the containerized application, configure env variables as follows.

docker run -e CONFIG_FORCE_topic=some-other-dlq --network=host kafka-dlq-monitor

Jar

To run create and run jar directly

./gradlew shadowJar
java -jar ./build/libs/kafka-dlq-monitor-1.0.0-all.jar

or with external config override:

./gradlew shadowJar
java -jar -Dconfig.file=./path/to/config.conf ./build/libs/kafka-dlq-monitor-1.0.0-all.jar