def execute(self, context: 'Context') -> List[str]:
emr_hook = EmrHook(aws_conn_id=self.aws_conn_id)
emr = emr_hook.get_conn()
job_flow_id = self.job_flow_id or emr_hook.get_cluster_id_by_name(
str(self.job_flow_name), self.cluster_states)
if not job_flow_id:
raise AirflowException(
f'No cluster found for name: {self.job_flow_name}')
if self.do_xcom_push:
context['ti'].xcom_push(key='job_flow_id', value=job_flow_id)
self.log.info('Adding steps to %s', job_flow_id)
# steps may arrive as a string representing a list
# e.g. if we used XCom or a file then: steps="[{ step1 }, { step2 }]"
steps = self.steps
if isinstance(steps, str):
steps = ast.literal_eval(steps)
response = emr.add_job_flow_steps(JobFlowId=job_flow_id, Steps=steps)
if not response['ResponseMetadata']['HTTPStatusCode'] == 200:
raise AirflowException(f'Adding steps failed: {response}')
else:
self.log.info('Steps %s added to JobFlow', response['StepIds'])
return response['StepIds']
def execute(self, context: 'Context') -> int:
emr_hook = EmrHook(aws_conn_id=self.aws_conn_id)
emr = emr_hook.get_conn()
if self.do_xcom_push:
context['ti'].xcom_push(key='cluster_id', value=self.cluster_id)
EmrClusterLink.persist(
context=context,
operator=self,
region_name=emr_hook.conn_region_name,
aws_partition=emr_hook.conn_partition,
job_flow_id=self.cluster_id,
)
self.log.info('Modifying cluster %s', self.cluster_id)
response = emr.modify_cluster(
ClusterId=self.cluster_id,
StepConcurrencyLevel=self.step_concurrency_level)
if response['ResponseMetadata']['HTTPStatusCode'] != 200:
raise AirflowException(f'Modify cluster failed: {response}')
else:
self.log.info('Steps concurrency level %d',
response['StepConcurrencyLevel'])
return response['StepConcurrencyLevel']
def execute(self, context: Dict[str, Any]) -> int:
emr_hook = EmrHook(aws_conn_id=self.aws_conn_id)
emr = emr_hook.get_conn()
if self.do_xcom_push:
context['ti'].xcom_push(key='cluster_id', value=self.cluster_id)
self.log.info('Modifying cluster %s', self.cluster_id)
response = emr.modify_cluster(
ClusterId=self.cluster_id, StepConcurrencyLevel=self.step_concurrency_level
)
if response['ResponseMetadata']['HTTPStatusCode'] != 200:
raise AirflowException(f'Modify cluster failed: {response}')
else:
self.log.info('Steps concurrency level %d', response['StepConcurrencyLevel'])
return response['StepConcurrencyLevel']
def execute(self, context: 'Context') -> None:
emr_hook = EmrHook(aws_conn_id=self.aws_conn_id)
emr = emr_hook.get_conn()
EmrClusterLink.persist(
context=context,
operator=self,
region_name=emr_hook.conn_region_name,
aws_partition=emr_hook.conn_partition,
job_flow_id=self.job_flow_id,
)
self.log.info('Terminating JobFlow %s', self.job_flow_id)
response = emr.terminate_job_flows(JobFlowIds=[self.job_flow_id])
if not response['ResponseMetadata']['HTTPStatusCode'] == 200:
raise AirflowException(f'JobFlow termination failed: {response}')
else:
self.log.info('JobFlow with id %s terminated', self.job_flow_id)
def execute(self, context):
emr_hook = EmrHook(aws_conn_id=self.aws_conn_id)
emr = emr_hook.get_conn()
job_flow_id = self.job_flow_id or emr_hook.get_cluster_id_by_name(
self.job_flow_name, self.cluster_states)
if not job_flow_id:
raise AirflowException(
f'No cluster found for name: {self.job_flow_name}')
if self.do_xcom_push:
context['ti'].xcom_push(key='job_flow_id', value=job_flow_id)
self.log.info('Adding steps to %s', job_flow_id)
response = emr.add_job_flow_steps(JobFlowId=job_flow_id,
Steps=self.steps)
if not response['ResponseMetadata']['HTTPStatusCode'] == 200:
raise AirflowException('Adding steps failed: %s' % response)
else:
self.log.info('Steps %s added to JobFlow', response['StepIds'])
return response['StepIds']
def test_get_conn_returns_a_boto3_connection(self):
hook = EmrHook(aws_conn_id='aws_default', region_name='ap-southeast-2')
self.assertIsNotNone(hook.get_conn().list_clusters())
