EMR automatically finishing when manually finishing is selected via Airflow

0

So I define manually finishing using the RunJobFlow operator (https://docs.aws.amazon.com/emr/latest/APIReference/API_RunJobFlow.html) "KeepJobFlowAliveWhenNoSteps": True. However, the cluster cluster automatically finishes but says terminated by user request.

EMR config

Terminated by user request, which is not true

This is the configuration I have defined in airflow:

JOB_FLOW_OVERRIDES = {
    "Name": "X",
    "Applications": [
        {"Name": "Spark"},
        {"Name": "Ganglia"},
        ],
    "Instances": {
        "KeepJobFlowAliveWhenNoSteps": True,
        "TerminationProtected": False,
        "Ec2SubnetId":"X",
        "EmrManagedMasterSecurityGroup": "X",
        "EmrManagedSlaveSecurityGroup": "X",
        "ServiceAccessSecurityGroup": "X",
        "InstanceGroups": [
            {
                "Name": "Master nodes",
                "Market": "SPOT",
                "InstanceRole": "MASTER",
                "InstanceType": "r4.xlarge",
                "InstanceCount": 1,
            },
            {
                "Name": "Core nodes",
                "Market": "SPOT",
                "InstanceRole": "CORE",
                "InstanceType": "r4.xlarge",
                "InstanceCount": 2,
            }
        ],
        "Ec2KeyName": "X",
    },
    "LogUri": "X",
    "ReleaseLabel": "X",
    "VisibleToAllUsers": True,
    "JobFlowRole": "DataPipelineDefaultResourceRole",
    "ServiceRole": "DataPipelineDefaultRole",
}
preguntada hace 5 meses151 visualizaciones
1 Respuesta
3

Hello,

Please check the CloudTrail and see which user initiates the TerminateJobFlows API call. There could be the user or the Org policy who gain access to terminate the cluster would be performed this. For an instance, if any policy that created in your Org when the cluster created using public subnet will be terminated automatically. You can get more insights from referring the CloudTrail. If the TerminateJobFlows triggered by Airflow role, then it is handled as part of DAG execution.

AWS
INGENIERO DE SOPORTE
respondido hace 5 meses
profile picture
EXPERTO
revisado hace 5 meses

No has iniciado sesión. Iniciar sesión para publicar una respuesta.

Una buena respuesta responde claramente a la pregunta, proporciona comentarios constructivos y fomenta el crecimiento profesional en la persona que hace la pregunta.

Pautas para responder preguntas