Beispiel #1
0
API_KEY = os.getenv("CDSW_API_KEY") 
PROJECT_NAME = os.getenv("CDSW_PROJECT")  

# Instantiate API Wrapper
cml = CMLBootstrap(HOST, USERNAME, API_KEY, PROJECT_NAME)

# Set the STORAGE environment variable
try : 
  storage=os.environ["STORAGE"]
except:
  tree = ET.parse('/etc/hadoop/conf/hive-site.xml')
  root = tree.getroot()
    
  for prop in root.findall('property'):
    if prop.find('name').text == "hive.metastore.warehouse.dir":
        storage = prop.find('value').text.split("/")[0] + "//" + prop.find('value').text.split("/")[2]
  storage_environment_params = {"STORAGE":storage}
  storage_environment = cml.create_environment_variable(storage_environment_params)
  os.environ["STORAGE"] = storage


if os.environ["STORAGE"] == "//tablespace":
    os.environ["STORAGE"] = "/tmp"

# Upload the data to the cloud storage
!hdfs dfs -mkdir -p $STORAGE/datalake
!hdfs dfs -mkdir -p $STORAGE/datalake/data
!hdfs dfs -mkdir -p $STORAGE/datalake/data/churn
!hdfs dfs -copyFromLocal /home/cdsw/raw/WA_Fn-UseC_-Telco-Customer-Churn-.csv $STORAGE/datalake/data/churn/WA_Fn-UseC_-Telco-Customer-Churn-.csv

Beispiel #2
0
    storage = os.environ["STORAGE"]
except:
    if os.path.exists("/etc/hadoop/conf/hive-site.xml"):
        tree = ET.parse("/etc/hadoop/conf/hive-site.xml")
        root = tree.getroot()
        for prop in root.findall("property"):
            if prop.find("name").text == "hive.metastore.warehouse.dir":
                storage = (
                    prop.find("value").text.split("/")[0]
                    + "//"
                    + prop.find("value").text.split("/")[2]
                )
    else:
        storage = "/user/" + os.getenv("HADOOP_USER_NAME")
    storage_environment_params = {"STORAGE": storage}
    storage_environment = cml.create_environment_variable(storage_environment_params)
    os.environ["STORAGE"] = storage

# define a function to run commands on HDFS
def run_cmd(cmd, raise_err=True):

    """
    Run Linux commands using Python's subprocess module

    Args:
        cmd (str) - Linux command to run
    Returns:
        process
    """
    print("Running system command: {0}".format(cmd))