Deployment
There are several ways to deploy the system, corresponding with the several artifacts that it produces.
Document Structure
Executable Jars
Requires:
- Java 8+
- A running kafka cluster
- A running ES cluster
You can deploy the components by setting env variables to point to the require config values and running the -all jars with java:
export KAFKA_BOOTSTRAP_SERVERS=...
java -jar registry/build/libs/psi-registry-$VERSION-all.jar
java -jar parsalyzer/build/libs/psi-parsalyzer-$VERSION-all.jar
...
Variation: External Servlet Container
The web application can also be deployed into an externally-managed servlet container using its war artifact. For example it might look something like this for a Tomcat installation:
cat << EOF > $CATALINA_HOME/bin/setenv.sh
#!/bin/bash
export KAFKA_BOOTSTRAP_SERVERS=...
EOF
cp registry/build/libs/psi-registry-$VERSION.war $CATALINA_HOME/webapps
$CATALINA_HOME/bin/shutdown.sh && $CATALINA_HOME/bin/startup.sh
Docker Images
Requires:
- Docker (or another container engine)
Another option is to use the docker images built by this project. For example, these commands will create and run containers for the registry and parsalyzer, as well as Kafka and Zookeeper using the well-maintained Confluent Platform images.
docker network create psi
docker run -d \
--name zookeeper \
--env ZOOKEEPER_CLIENT_PORT=2181 \
--network psi \
--expose 2181 \
confluentinc/cp-zookeeper
docker run -d \
--name kafka \
--env KAFKA_ZOOKEEPER_CONNECT=zookeeper:2181 \
--env KAFKA_ADVERTISED_LISTENERS=PLAINTEXT://kafka:9092 \
--env KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR=1 \
--network psi \
--expose 9092 \
confluentinc/cp-kafka
docker run -d \
--name schema-registry \
--env SCHEMA_REGISTRY_KAFKASTORE_BOOTSTRAP_SERVERS=PLAINTEXT://kafka:9092 \
--env SCHEMA_REGISTRY_HOST_NAME=schema-registry \
--network psi \
--expose 8081 \
confluentinc/cp-schema-registry
docker run -d \
--name psi-registry \
--env KAFKA_BOOTSTRAP_SERVERS=kafka:9092 \
--env SCHEMA_REGISTRY_URL=http://schema-registry:8081 \
--network psi \
-p 8080 \
cedardevs/psi-registry
docker run -d \
--name psi-manager \
--env KAFKA_BOOTSTRAP_SERVERS=kafka:9092 \
--env SCHEMA_REGISTRY_URL=schema-registry:8081 \
--network psi \
cedardevs/psi-parsalyzer
Of course higher level tools can be used to manage these containers, like docker-compose, docker swarm, or…
Kubernetes and Helm
Requirements:
- kubectl, configured to point to any…
- Kubernetes cluster (e.g. minikube or docker desktop)
- The helm cli installed locally
- Tiller running in the cluster, via
helm init
Finally, this project contains a helm chart located in in the helm directory. To deploy the entire system simply run:
helm install ./helm/psi