Missing LogUri from emr describe-cluster API when executing EmrCreateJobFlowOperator

[Lee-W]

Apache Airflow version

main (development)

What happened

Encounter the following error when executing EmrCreateJobFlowOperator

Traceback (most recent call last):
  File "/usr/local/lib/python3.10/site-packages/airflow/providers/amazon/aws/operators/emr.py", line 695, in execute
    log_uri=get_log_uri(emr_client=self._emr_hook.conn, job_flow_id=self._job_flow_id),
  File "/usr/local/lib/python3.10/site-packages/airflow/providers/amazon/aws/links/emr.py", line 61, in get_log_uri
    log_uri = S3Hook.parse_s3_url(response["Cluster"]["LogUri"])
KeyError: 'LogUri'

According to the this document, it seems we might not always be able to get ["Cluster"]["LogUri"] and we encounter errors after the release in #31322.

What you think should happen instead

The EmrCreateJobFlowOperator should finish execution without error.

How to reproduce

1. git clone github.com/apache/airflow/
2. cd airflow
3. git checkout c082aec089405ed0399cfee548011b0520be0011 (the main branch when I found this issue)
4. Add the following DAG to files/dags/ and name it as example_emr.py

import os
from datetime import datetime, timedelta

from airflow import DAG
from airflow.providers.amazon.aws.operators.emr import EmrCreateJobFlowOperator, EmrTerminateJobFlowOperator


JOB_FLOW_OVERRIDES = {
    "Name": "example_emr_sensor_cluster",
    "ReleaseLabel": "emr-5.29.0",
    "Applications": [{"Name": "Spark"}],
    "Instances": {
        "InstanceGroups": [
            {
                "Name": "Primary node",
                "Market": "ON_DEMAND",
                "InstanceRole": "MASTER",
                "InstanceType": "m4.large",
                "InstanceCount": 1,
            },
        ],
        "KeepJobFlowAliveWhenNoSteps": False,
        "TerminationProtected": False,
    },
    "JobFlowRole": "EMR_EC2_DefaultRole"
    "ServiceRole": "EMR_DefaultRole",
}

DEFAULT_ARGS = {
    "execution_timeout": timedelta(hours=6),
    "retries": 2,
    "retry_delay": 60,
}


with DAG(
    dag_id="example_emr_sensor",
    schedule=None,
    start_date=datetime(2022, 1, 1),
    default_args=DEFAULT_ARGS,
    catchup=False,
) as dag:
    create_job_flow = EmrCreateJobFlowOperator(
        task_id="create_job_flow",
        job_flow_overrides=JOB_FLOW_OVERRIDES,
        aws_conn_id="aws_default",
    )

    remove_job_flow = EmrTerminateJobFlowOperator(
        task_id="remove_job_flow",
        job_flow_id=create_job_flow.output,
        aws_conn_id="aws_default",
        trigger_rule="all_done",
    )

    create_job_flow >> remove_job_flow

5. breeze --python 3.8 --backend sqlite start-airflow
6. Trigger the DAG from web UI

Operating System

mac OS 13.4

Versions of Apache Airflow Providers

No response

Deployment

Astronomer

Deployment details

No response

Anything else

#31322 (comment)

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[boring-cyborg]

Thanks for opening your first issue here! Be sure to follow the issue template! If you are willing to raise PR to address this issue please do so, no need to wait for approval.