Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add an Amazon EMR on EKS provider package #16766

Merged
merged 9 commits into from
Aug 27, 2021
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
81 changes: 81 additions & 0 deletions airflow/providers/amazon/aws/example_dags/example_emr_eks_job.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,81 @@
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
"""
This is an example dag for an Amazon EMR on EKS Spark job.
"""
import os
from datetime import timedelta

from airflow import DAG
from airflow.providers.amazon.aws.operators.emr_containers import EMRContainerOperator
from airflow.utils.dates import days_ago

# [START howto_operator_emr_eks_env_variables]
VIRTUAL_CLUSTER_ID = os.getenv("VIRTUAL_CLUSTER_ID", "test-cluster")
JOB_ROLE_ARN = os.getenv("JOB_ROLE_ARN", "arn:aws:iam::012345678912:role/emr_eks_default_role")
# [END howto_operator_emr_eks_env_variables]


# [START howto_operator_emr_eks_config]
JOB_DRIVER_ARG = {
"sparkSubmitJobDriver": {
"entryPoint": "local:///usr/lib/spark/examples/src/main/python/pi.py",
"sparkSubmitParameters": "--conf spark.executors.instances=2 --conf spark.executors.memory=2G --conf spark.executor.cores=2 --conf spark.driver.cores=1", # noqa: E501
}
}

CONFIGURATION_OVERRIDES_ARG = {
"applicationConfiguration": [
{
"classification": "spark-defaults",
"properties": {
"spark.hadoop.hive.metastore.client.factory.class": "com.amazonaws.glue.catalog.metastore.AWSGlueDataCatalogHiveClientFactory", # noqa: E501
},
}
],
"monitoringConfiguration": {
"cloudWatchMonitoringConfiguration": {
"logGroupName": "/aws/emr-eks-spark",
"logStreamNamePrefix": "airflow",
}
},
}
# [END howto_operator_emr_eks_config]

with DAG(
dag_id='emr_eks_pi_job',
dagrun_timeout=timedelta(hours=2),
start_date=days_ago(1),
schedule_interval="@once",
tags=["emr_containers", "example"],
) as dag:

# An example of how to get the cluster id and arn from an Airflow connection
# VIRTUAL_CLUSTER_ID = '{{ conn.emr_eks.extra_dejson["virtual_cluster_id"] }}'
# JOB_ROLE_ARN = '{{ conn.emr_eks.extra_dejson["job_role_arn"] }}'

# [START howto_operator_emr_eks_jobrun]
job_starter = EMRContainerOperator(
task_id="start_job",
virtual_cluster_id=VIRTUAL_CLUSTER_ID,
execution_role_arn=JOB_ROLE_ARN,
release_label="emr-6.3.0-latest",
job_driver=JOB_DRIVER_ARG,
configuration_overrides=CONFIGURATION_OVERRIDES_ARG,
name="pi.py",
)
# [END howto_operator_emr_eks_jobrun]
205 changes: 205 additions & 0 deletions airflow/providers/amazon/aws/hooks/emr_containers.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,205 @@
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.

from time import sleep
from typing import Any, Dict, Optional

from botocore.exceptions import ClientError

from airflow.exceptions import AirflowException
from airflow.providers.amazon.aws.hooks.base_aws import AwsBaseHook


class EMRContainerHook(AwsBaseHook):
"""
Interact with AWS EMR Virtual Cluster to run, poll jobs and return job status
Additional arguments (such as ``aws_conn_id``) may be specified and
are passed down to the underlying AwsBaseHook.

.. seealso::
:class:`~airflow.providers.amazon.aws.hooks.base_aws.AwsBaseHook`

:param virtual_cluster_id: Cluster ID of the EMR on EKS virtual cluster
:type virtual_cluster_id: str
"""

INTERMEDIATE_STATES = (
"PENDING",
"SUBMITTED",
"RUNNING",
)
FAILURE_STATES = (
"FAILED",
"CANCELLED",
"CANCEL_PENDING",
)
SUCCESS_STATES = ("COMPLETED",)

def __init__(self, *args: Any, virtual_cluster_id: str = None, **kwargs: Any) -> None:
super().__init__(client_type="emr-containers", *args, **kwargs) # type: ignore
self.virtual_cluster_id = virtual_cluster_id

def submit_job(
self,
name: str,
execution_role_arn: str,
release_label: str,
job_driver: dict,
configuration_overrides: Optional[dict] = None,
client_request_token: Optional[str] = None,
) -> str:
"""
Submit a job to the EMR Containers API and and return the job ID.
A job run is a unit of work, such as a Spark jar, PySpark script,
or SparkSQL query, that you submit to Amazon EMR on EKS.
See: https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/emr-containers.html#EMRContainers.Client.start_job_run # noqa: E501

:param name: The name of the job run.
:type name: str
:param execution_role_arn: The IAM role ARN associated with the job run.
:type execution_role_arn: str
:param release_label: The Amazon EMR release version to use for the job run.
:type release_label: str
:param job_driver: Job configuration details, e.g. the Spark job parameters.
:type job_driver: dict
:param configuration_overrides: The configuration overrides for the job run,
specifically either application configuration or monitoring configuration.
:type configuration_overrides: dict
:param client_request_token: The client idempotency token of the job run request.
Use this if you want to specify a unique ID to prevent two jobs from getting started.
:type client_request_token: str
:return: Job ID
"""
params = {
"name": name,
"virtualClusterId": self.virtual_cluster_id,
"executionRoleArn": execution_role_arn,
"releaseLabel": release_label,
"jobDriver": job_driver,
"configurationOverrides": configuration_overrides or {},
}
if client_request_token:
params["clientToken"] = client_request_token

response = self.conn.start_job_run(**params)

if response['ResponseMetadata']['HTTPStatusCode'] != 200:
raise AirflowException(f'Start Job Run failed: {response}')
else:
self.log.info(
"Start Job Run success - Job Id %s and virtual cluster id %s",
response['id'],
response['virtualClusterId'],
)
return response['id']

def get_job_failure_reason(self, job_id: str) -> Optional[str]:
"""
Fetch the reason for a job failure (e.g. error message). Returns None or reason string.

:param job_id: Id of submitted job run
:type job_id: str
:return: str
"""
# We absorb any errors if we can't retrieve the job status
reason = None

try:
response = self.conn.describe_job_run(
virtualClusterId=self.virtual_cluster_id,
id=job_id,
)
reason = response['jobRun']['failureReason']
except KeyError:
self.log.error('Could not get status of the EMR on EKS job')
except ClientError as ex:
self.log.error('AWS request failed, check logs for more info: %s', ex)

return reason

def check_query_status(self, job_id: str) -> Optional[str]:
"""
Fetch the status of submitted job run. Returns None or one of valid query states.
See: https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/emr-containers.html#EMRContainers.Client.describe_job_run # noqa: E501
:param job_id: Id of submitted job run
:type job_id: str
:return: str
"""
try:
response = self.conn.describe_job_run(
virtualClusterId=self.virtual_cluster_id,
id=job_id,
)
return response["jobRun"]["state"]
except self.conn.exceptions.ResourceNotFoundException:
# If the job is not found, we raise an exception as something fatal has happened.
raise AirflowException(f'Job ID {job_id} not found on Virtual Cluster {self.virtual_cluster_id}')
except ClientError as ex:
# If we receive a generic ClientError, we swallow the exception so that the
self.log.error('AWS request failed, check logs for more info: %s', ex)
return None

def poll_query_status(
self, job_id: str, max_tries: Optional[int] = None, poll_interval: int = 30
) -> Optional[str]:
"""
Poll the status of submitted job run until query state reaches final state.
Returns one of the final states.

:param job_id: Id of submitted job run
:type job_id: str
:param max_tries: Number of times to poll for query state before function exits
:type max_tries: int
:param poll_interval: Time (in seconds) to wait between calls to check query status on EMR
:type poll_interval: int
:return: str
"""
try_number = 1
final_query_state = None # Query state when query reaches final state or max_tries reached

# TODO: Make this logic a little bit more robust.
# Currently this polls until the state is *not* one of the INTERMEDIATE_STATES
# While that should work in most cases...it might not. :)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you tell me a little more about it?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

After thinking about this a little bit more, I think my concern was more about the logic here solely relying on the INTERMEDIATE_STATES.

What that means is if the API ever changes (not likely), the logic here could break. I think the only change I would make here would be a more explicit check if the query_state is actually in a completed state...but that's the current logic anyway because there's either None state, INTERMEDIATE state, or COMPLETED state.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Opened followup task #19877

while True:
query_state = self.check_query_status(job_id)
if query_state is None:
self.log.info("Try %s: Invalid query state. Retrying again", try_number)
elif query_state in self.INTERMEDIATE_STATES:
self.log.info("Try %s: Query is still in an intermediate state - %s", try_number, query_state)
else:
self.log.info("Try %s: Query execution completed. Final state is %s", try_number, query_state)
final_query_state = query_state
break
if max_tries and try_number >= max_tries: # Break loop if max_tries reached
final_query_state = query_state
break
try_number += 1
sleep(poll_interval)
return final_query_state
Comment on lines +177 to +192

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Have you thought about jobs that are not expected to terminate in a relatively short time? If I submit a streaming job for spark using this operator, then my job needs to be running for a longer time. Do you need to implement some kind of backoff-algorithm based checks on this?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@wanderijames That's a good point. I don't know if we need to implement a backoff, we do already have the option to change the poll interval. But that's also where I think your PR is nice in that it has the operator to just start the job.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cool. Thanks


def stop_query(self, job_id: str) -> Dict:
"""
Cancel the submitted job_run

:param job_id: Id of submitted job_run
:type job_id: str
:return: dict
"""
return self.conn.cancel_job_run(
virtualClusterId=self.virtual_cluster_id,
id=job_id,
)
Loading