先試後買

A. V2 uses manifest lists instead of manifest files for tracking data files.

B. V2 supports new data types like UUIDs, which are unavailable in V1.

C. V2 tables are generally less performant than V1 tables due to added metadata overhead.

D. V2 introduces mandatory partitioning, while V1 allows for unpartitioned tables.

A. Manually document data lineage within the YAML configuration files.

B. Leverage the integrated lineage tracking capabilities of the Cloudera Data Engineering service.

C. Integrate the service with a separate third-party lineage tracking tool.

D. Utilize custom scripting to track data movement between pipeline steps.

A. All of the above

B. Insufficient disk space on the cluster nodes

C. Inefficient serialization format for the RDD data

D. Using HDFS as the storage location instead of local disks

A. Implement extensive logging within each task, regardless of its purpose, to capture detailed execution information.

B. Break down the DAG into smaller sub-DAGs with well-defined functionalities and clear naming conventions.

C. Utilize comments sparingly within the DAG code, as the logic should be self-explanatory.

D. Use cryptic and abbreviated names for tasks and variables, assuming everyone understands the context.

A. It reduces the need for data compression

B. It provides metadata for security enforcement

C. It is used to generate more data

D. It helps the optimizer make informed decisions about data layout and query execution plans

A. Use the EmailOperator to send email notifications with details about the task status.

B. Implement custom logic within each task to send notifications using external libraries.

C. Configure the Airflow web Ul to send alerts based on task execution status.

D. Utilize Airflow variables to store notification details and access them within tasks.

A. Too many tasks assigned to each Executor.

B. Insufficient memory allocation for Executor pods.

C. Executor pods are not configured to restart upon failure.

D. Dynamic resource allocation is disabled.

A. By storing all data on a single node

B. By transferring only required data between executors

C. Spark doesn't perform data shuffling

D. By broadcasting all data to each executor

A. Loop through the DataFrame and group purchases by month, calculating the average manually

B. Convert the "purchase_date" column to a string and filter based on the year

C. Utilize Spark's machine learning library (MLIiB. for time series analysis

D. Use Spark SQL's MONTH function and AVG function with appropriate windowing

A. Manually trigger the dependent DAG after the first DAG completes.

B. Use the TriggerDagRunOperator to trigger one DAG from another upon completion.

C. Directly call one DAG from another using the PythonOperator.

D. Use the ExternalTaskSensor to wait for a task in another DAG to complete.

A. All of the above

B. The Airflow web UI task details page might display error messages related to the failed check.

C. You can directly inspect the custom Python script used for the check to identify potential issues.

D. The Airflow logs (scheduler, worker) might contain relevant error messages from the failing task.

A. Package the libraries with your DAG code and reference them within the Python operators.

B. Configure Airflow to use a virtual environment with pre-installed libraries.

C. Install the required libraries directly into the Airflow environment.

D. Utilize system-wide library installations, assuming they are accessible to the Airflow user.