pyspark and HDFS commands

You can delete an hdfs path in PySpark without using third party dependencies as follows:

from pyspark.sql import SparkSession
# example of preparing a spark session
spark = SparkSession.builder.appName('abc').getOrCreate()
sc = spark.sparkContext
# Prepare a FileSystem manager
fs = (sc._jvm.org
      .apache.hadoop
      .fs.FileSystem
      .get(sc._jsc.hadoopConfiguration())
      )
path = "Your/hdfs/path"
# use the FileSystem manager to remove the path
fs.delete(sc._jvm.org.apache.hadoop.fs.Path(path), True)

---

To improve one step further, you can wrap the above idea into a helper function that you can re-use across jobs/packages:

from pyspark.sql import SparkSession
spark = SparkSession.builder.appName('abc').getOrCreate()

def delete_path(spark, path):
    sc = spark.sparkContext
    fs = (sc._jvm.org
          .apache.hadoop
          .fs.FileSystem
          .get(sc._jsc.hadoopConfiguration())
          )
    fs.delete(sc._jvm.org.apache.hadoop.fs.Path(path), True)

delete_path(spark, "Your/hdfs/path")

from https://diogoalexandrefranco.github.io/interacting-with-hdfs-from-pyspark/ using only PySpark

######
# Get fs handler from java gateway
######
URI = sc._gateway.jvm.java.net.URI
Path = sc._gateway.jvm.org.apache.hadoop.fs.Path
FileSystem = sc._gateway.jvm.org.apache.hadoop.fs.FileSystem
fs = FileSystem.get(URI("hdfs://somehost:8020"), sc._jsc.hadoopConfiguration())

# We can now use the Hadoop FileSystem API (https://hadoop.apache.org/docs/current/api/org/apache/hadoop/fs/FileSystem.html)
fs.listStatus(Path('/user/hive/warehouse'))
# or
fs.delete(Path('some_path'))

the other solutions didn't work in my case, but this blog post helped :)

You can execute arbitrary shell command using form example subprocess.call or sh library so something like this should work just fine:

import subprocess

some_path = ...
subprocess.call(["hadoop", "fs", "-rm", "-f", some_path])

If you use Python 2.x you can try using spotify/snakebite:

from snakebite.client import Client

host = ...
port = ...
client = Client(host, port)
client.delete(some_path, recurse=True)

hdfs3 is yet another library which can be used to do the same thing:

from hdfs3 import HDFileSystem

hdfs = HDFileSystem(host=host, port=port)
HDFileSystem.rm(some_path)

Apache Arrow Python bindings are the latest option (and that often is already available on Spark cluster, as it is required for pandas_udf):

from pyarrow import hdfs

fs = hdfs.connect(host, port)
fs.delete(some_path, recursive=True)