Amazon EMR on EKS용 Flink 네이티브 Kubernetes 시작하기

Flink 네이티브 Kubernetes 명령으로 Flink 애플리케이션을 실행하려면 먼저 Amazon EMR on EKS에서 Flink 네이티브 Kubernetes 설정의 단계를 완료합니다.

Flink 다운로드 및 설치.

다음과 같은 환경 변수의 값을 설정합니다.

#Export the FLINK_HOME environment variable to your local installation of Flink
export FLINK_HOME=/usr/local/bin/flink #Will vary depending on your installation
export NAMESPACE=flink
export CLUSTER_ID=flink-application-cluster
export IMAGE=<123456789012.dkr.ecr.sample-AWS 리전-.amazonaws.com/flink/emr-6.13.0-flink:latest>
export FLINK_SERVICE_ACCOUNT=emr-containers-sa-flink
export FLINK_CLUSTER_ROLE_BINDING=emr-containers-crb-flink

Kubernetes 리소스를 관리할 서비스 계정을 생성합니다.

kubectl create serviceaccount $FLINK_SERVICE_ACCOUNT -n $NAMESPACE
kubectl create clusterrolebinding $FLINK_CLUSTER_ROLE_BINDING --clusterrole=edit --serviceaccount=$NAMESPACE:$FLINK_SERVICE_ACCOUNT

run-application CLI 명령을 실행합니다.

$FLINK_HOME/bin/flink run-application \
    --target kubernetes-application \
    -Dkubernetes.namespace=$NAMESPACE \
    -Dkubernetes.cluster-id=$CLUSTER_ID \
    -Dkubernetes.container.image.ref=$IMAGE \
    -Dkubernetes.service-account=$FLINK_SERVICE_ACCOUNT \
    local:///opt/flink/examples/streaming/Iteration.jar
2022-12-29 21:13:06,947 INFO  org.apache.flink.kubernetes.utils.KubernetesUtils            [] - Kubernetes deployment requires a fixed port. Configuration blob.server.port will be set to 6124
2022-12-29 21:13:06,948 INFO  org.apache.flink.kubernetes.utils.KubernetesUtils            [] - Kubernetes deployment requires a fixed port. Configuration taskmanager.rpc.port will be set to 6122
2022-12-29 21:13:07,861 WARN  org.apache.flink.kubernetes.KubernetesClusterDescriptor      [] - Please note that Flink client operations(e.g. cancel, list, stop, savepoint, etc.) won't work from outside the Kubernetes cluster since 'kubernetes.rest-service.exposed.type' has been set to ClusterIP.
2022-12-29 21:13:07,868 INFO  org.apache.flink.kubernetes.KubernetesClusterDescriptor      [] - Create flink application cluster flink-application-cluster successfully, JobManager Web Interface: http://flink-application-cluster-rest.flink:8081
                 

생성된 Kubernetes 리소스를 검사합니다.

kubectl get all -n <namespace>
NAME READY STATUS RESTARTS AGE
pod/flink-application-cluster-546687cb47-w2p2z 1/1 Running 0 3m37s
pod/flink-application-cluster-taskmanager-1-1 1/1 Running 0 3m24s

NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/flink-application-cluster ClusterIP None <none> 6123/TCP,6124/TCP 3m38s
service/flink-application-cluster-rest ClusterIP 10.100.132.158 <none> 8081/TCP 3m38s

NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/flink-application-cluster 1/1 1 1 3m38s

NAME DESIRED CURRENT READY AGE
replicaset.apps/flink-application-cluster-546687cb47 1 1 1 3m38s
                     

포트는 8081로 포워딩됩니다.

kubectl port-forward service/flink-application-cluster-rest 8081 -n <namespace>
Forwarding from 127.0.0.1:8081 -> 8081

Flink UI에 로컬로 액세스합니다.

Flink 애플리케이션을 삭제합니다.

kubectl delete deployment.apps/flink-application-cluster -n <namespace>
deployment.apps "flink-application-cluster" deleted