First of all run the command to start Minio
docker run -e MINIO_ROOT_USER=minio -e MINIO_ROOT_PASSWORD=miniostorage --net=host minio/minio server /data --console-address ":9001"
In the routes.yaml file, set correctly the Minio credentials for your bucket.
Download the minio client too:
wget https://dl.min.io/client/mc/release/linux-amd64/mc chmod +x mc sudo mv mc /usr/local/bin
Now we need to set up an alias for our bucket and login
mc alias set minio http://127.0.0.1:9000 minio miniostorage mc mb minio/ckc
Now we can use the minio-bulk.sh script
./minio-bulk.sh -f msg1.txt -b ckc -n 10000
Also you'll need to run a Kafka cluster to point to. In this case you could use an ansible role like https://github.com/oscerd/kafka-ansible-role
And set up a file deploy.yaml with the following content:
- name: role kafka hosts: localhost remote_user: user roles: - role: kafka-ansible-role kafka_version: 2.8.0 path_dir: /home/user/ unarchive_dest_dir: /home/user/kafka/demo/ start_kafka: true
and then run
ansible-playbook -v deploy.yaml
This should start a Kafka instance for you, on your local machine.
build:
./mvnw clean package
If you want to have JFR enable from the beginning:
build:
./mvnw clean package -Pjfr
At this point you're able to run the example:
docker:
docker run --rm -ti \ -v $PWD/data:/etc/camel:Z \ -e CAMEL_K_CONF=/etc/camel/application.properties \ --network="host" \ quay.io/oscerd/minio-kafka:1.0-SNAPSHOT-jvm
docker:
docker run --rm -ti \ -v $PWD/data:/etc/camel:Z \ -v $PWD/jfr:/work/jfr:Z \ -e CAMEL_K_CONF=/etc/camel/application.properties \ --network="host" \ quay.io/oscerd/minio-kafka:1.0-SNAPSHOT-jvm
Now you can start JFR with the following command
docker exec -it <container_id> jcmd 1 JFR.start name=Test settings=jfr/settings_for_heap.jfc duration=5m filename=jfr/output.jfr
and check the status
docker exec -it <container_id> jcmd 1 JFR.check
docker:
docker run --rm -ti \ -v $PWD/data:/etc/camel:Z \ -v async_profiler_path:/work/async-profiler:Z \ -e CAMEL_K_CONF=/etc/camel/application.properties \ --network="host" \ quay.io/oscerd/minio-kafka:1.0-SNAPSHOT-jvm
Where async profiler path is the path of your async profiler on your host machine.
Now you can start Async Profiler with the following command
docker exec -it <container_id> /work/async-profiler/profiler.sh -e alloc -d 30 -f /work/async-profiler/alloc_profile.html 1
This command while create an allocation flamegraph for the duration of 30 second of the running application.
The privileged option for running the docker container is the fastest way to have perf events syscall enabled.
If you don't want to use privileged approach, you can have a look at the basic configuration of async profiler (https://github.com/jvm-profiling-tools/async-profiler/wiki/Basic-Usage)
You could also modify the resources of your container with memory and cpu defined while running it
docker:
docker run --rm -ti \ -v $PWD/data:/etc/camel:Z \ -v $PWD/jfr:/work/jfr:Z \ -e CAMEL_K_CONF=/etc/camel/application.properties \ --network="host" \ -m 128m \ --cpu-quota="25000" \ quay.io/oscerd/minio-kafka:1.0-SNAPSHOT-jvm
In this case we are allocating 128 Mb Memory to the container and 0.25% cpus.
In the pom you can also set a different Heap Size. The default is 64 Mb.