Kafka to Solr

In the routes.yaml file, you'll need to set correctly the configuration for Solr and Kafka.

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 package

docker:

docker run --rm -ti \
    -v $PWD/data:/etc/camel:Z \
    -e CAMEL_K_CONF=/etc/camel/application.properties \
    --network="host" \
    quay.io/oscerd/kafka-solr:1.0-SNAPSHOT-jvm

Solr instance

You will also need a Solr instance, you can either run it locally or use a docker image like:

docker run --name solr_demo -d -p 8983:8983 solr solr-demo

Enabling JFR

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/kafka-solr:1.0-SNAPSHOT-jvm

You'll need a running Kafka broker locally on your host.

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

Enabling Async Profiler while running application

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/kafka-solr: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)

Tuning Container

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/kafka-solr:1.0-SNAPSHOT-jvm

In this case we are allocating 128 Mb Memory to the container and 0.25% cpus.

Send messages to Kafka

You'll need also kafkacat to be able to inject the filename header and use the burst script

export KAFKACAT_PATH=<path_to_your_kafkacat>

And now run the burst script.

This command for example will send 1000 messages with payload “payload” to the topic “testtopic”

cd script/
> ./burst.sh -b localhost:9092 -n 1000 -t testtopic -p '{ "id": "9", "title": [ "Doc 1", "Doc 2", "Doc 3" ], "genre": "Sci-fi" }'

You could also tests this approach with multiple producers, through the multiburst script

cd script/
> ./multiburst.sh -s 5 -b localhost:9092 -n 1000 -t testtopic -p '{ "id": "9", "title": [ "Doc 1", "Doc 2", "Doc 3" ], "genre": "Sci-fi" }'

This command will run 5 burst script with 1000 messages each one with payload “payload” to the Kafka instance running on localhost:9092 and the topic “testtopic”

You can check the documents in solr console from your docker image or local instance.s