This image contains the official Kafka binaries with focus on running command line tools for working with Kafka topics.

A secondary goal is to minimize the image size.

We face since years the bad practice of docker images permanently increasing in size with even simple requirements for using tools with simple scripts.

With Kafka, I don't want to have a pure Kafka client tools image based on the kafka server images like from confluent reaching 1 GB in size. They contain everything we need, but 1 GB for calling a script to list topics is unacceptable.

On the other hand, I wanted to be still with Java, so I don't care about alternative solutions (right now), sorry ;-)

The minimal image size of 145 MB was reachable with an OpenJDK base image plus some Kafka cleanup:

docker build -f Dockerfile.openjdk-jre-alpine -t jforge/kafka-tools .

Override Java version for the Dockerfile based on alpine:

docker build -f Dockerfile.alpine --build-arg JAVA_VERSION=17 -t jforge/kafka-tools .

Kafka still supports Java 8 in Kafka 3.6, but this will disappear in future release. Still, the JRE 8 image is the smallest compared to newer JREs:

OpenJDK images are deprecated, but I prefer the JRE 8 due to the minimal image size. Switch to Amazon Corretto, if you prefer non-deprecated base images.

Further information:

• I haven't investigated the Java runtime images from ìbm-semeru-runtimes, imbjava, sapmachine any further (they seem all to be bigger).
• For eclipse-temurin Java runtime images a simple test shows heap space errors, which seem to be resolvable with adding the security option --security-opt seccomp=unconfined to the docker run call. See Seccomp security profiles for Docker for details. For me, this adds unacceptable user-facing complexity, therefore skipped.
• The alternative which includes the Python kafka-tools is not efficiently created and too big, I left it here for some special case.
• I'm aware of, but didn't follow the vulnerability information of docker scout.
• The applied cleanup includes removing a very big jar file for a Kafka Streams dependency to RocksDB. So far, I didn't see a toxic effect. Please notify me in case, you get into trouble with this.

docker run --rm jforge/kafka-tools kafka-topics.sh <args>

Use the test.sh script to check the tools with a kafka cluster, e.g.:

test.sh --bootstrap-server <hostname>:9092 -p 3

For a test cluster use the single node docker composition (change the kafka advertised listener address accordingly):

docker compose -f docker-compose-kafka-simple.yaml up -d

docker run --rm jforge/kafka-tools kafka-topics.sh \
    --bootstrap-server localhost:9092 \
    --create --topic sample-topic \
    --partitions 20 --replication-factor 3

docker run --rm jforge/kafka-tools kafka-topics.sh \
    --bootstrap-server localhost:9092 \
    --describe sample-topic

docker run --rm jforge/kafka-tools kafka-topics.sh \
    --bootstrap-server localhost:9092 \
    --list

docker run --rm -it jforge/kafka-tools kafka-console-producer.sh \
    --bootstrap-server localhost:9092 \
    --topic sample-topic

docker run --rm -it jforge/kafka-tools kafka-console-consumer.sh \
  --bootstrap-server localhost:9092 \
  --topic sample-topic --from-beginning

• Kafka Basic operations
• Kafka Java prerequisites
• Docker Hub Amazon Corretto
• Docker Hub OpenJDK