DataFlow失败，返回代码1与气流DataflowHook.start_python_dataflow

Question

下面是我的密码。

当我运行下面的代码时，我得到了下面的错误。我正在尝试使用将google云存储中的gvcf/vcf文件转换为bigquery。

2018-06-06 16:46:42,589 {models.py:1428} INFO - 2018-06-06 21:46:34.252526 2018-06-06 16:46:42,589 {base_task_runner.py:115} INFO -运行：'bash'，'-c'，U‘’airflow运行GcsToBigQuery gcsToBigquery_ID 2018-06-06T21:46:34.252526 -job_id 168 -原始-sd DAGS_base_task_runner.py:98/gcsToBigQuery.py‘{base_task_runner.py:98} INFO -子任务: 2018-06-06 16:46:43 202 {init.py:45} INFO -使用执行者SequentialExecutor 2018-06-06 16:46:43,284 {base_task_runner.py:98} INFO -子任务:2018-06 16:46:43,283 {models.py：189) INFO -从/apps/气流/dags/GcsToBigQuery.py 2018-06-06 16:46:43,853 {base_task_runner.py:98} INFO -子任务: 2018-06-06 16:46:43,852 {gcp_dataflow_hook.py:111} INFO -开始等待DataFlow进程完成。2018-06 16:46:46:46:46,931 {base_task_runner.py:98} INFO -子任务: 2018-06-06 16:46:46:46,930 {GcsToBigQuery.py:48}错误-状态:失败: gcsToBigquery:无法运行：DataFlow失败，返回代码1 2018-06-06 16:46:46,931 {base_task_runner.py:98} INFO -子任务: 2018-06-06 16:46:46,930 {python_operator.py:90} INFO -完成。返回值为:无

请帮我解决这个问题。谢谢!

from datetime import datetime, timedelta
from airflow import DAG
from airflow.contrib.hooks.gcp_dataflow_hook import DataFlowHook
from airflow.operators.python_operator import PythonOperator
import logging

default_args = {
    'owner': 'My Name',
    'depends_on_past': False,
    'start_date': datetime(2018, 6, 6),
    'email': ['MY Email'],
    'email_on_failure': True,
    'email_on_retry': False,
    'retries': 2,
    'retry_delay': timedelta(minutes=5)
}

dag = DAG('GcsToBigQuery', default_args=default_args,
          description='To move GVCF/VCF files from Google Cloud Storage to Big Query',
          schedule_interval='@once',
          start_date=datetime(2018, 6, 6))

dataflow_py_file = 'gcp_variant_transforms.vcf_to_bq'
PY_OPTIONS = ['-m']

DATAFLOW_OPTIONS_PY = {
    "project": "project-Name",
    "input_pattern": "gs://test-gvcf/1000-genomes.vcf",
    "output_table": "trc-mayo-projectsample:1000genomicsID.1000_genomesSamp",
     "staging_location": "gs://test-gvcf/vcftobq/staging",
     "temp_location": "gs://test-gvcf/vcftobq/temp",
     "job_name": "dataflowstarter25",
     #"setup_file": "./setup.py",
     "runner": "DataflowRunner"
}


def gcsToBigquery():
    try:
        dataflowHook = DataFlowHook(gcp_conn_id='google_cloud_platform_id')
        dataflowHook.start_python_dataflow(task_id='dataflowStarter2_ID', variables=DATAFLOW_OPTIONS_PY,
                                       dataflow=dataflow_py_file, py_options=PY_OPTIONS)
    except Exception as e:
        logging.error("Status : FAIL : gcsToBigquery: Not able to run: " + str(e.message))

gcsToBigquery_task = PythonOperator(task_id='gcsToBigquery_ID',
                                    python_callable=gcsToBigquery,
                                    dag=dag)

Answer 1

通过使用DataflowPythonOperator并在云实例中安装gcp_variant_transforms API，可以避免此问题。

安装所需组件的命令：

sudo pip install git+https://github.com/googlegenomics/gcp-variant-transforms.git

如果有人也有此问题，您可以查看这个帖子，其中详细说明了斯里坎成功执行代码所遵循的步骤。