使用架构详细信息创建数据框时 Dataproc 上的 Pyspark 错误

Posted 2023-04-15

【中文标题】使用架构详细信息创建数据框时 Dataproc 上的 Pyspark 错误

【英文标题】：Pyspark error on Dataproc while creating dataframe with the schema details

【发布时间】：2020-01-29 12:08:23

【问题描述】：

我有一个带有 Anaconda 的 Dataproc 集群。我创建了一个虚拟环境。在 anaconda my-env 内部，因为我需要在那里安装开源 RDkit，因此我再次安装了 PySpark（不使用预装的）。现在使用下面的代码，我在my-env 中出现错误，但在my-env 之外没有出现错误

代码：

from pyspark.sql.types import StructField, StructType, StringType, LongType
from pyspark.sql import SparkSession
from py4j.protocol import Py4JJavaError
spark = SparkSession.builder.appName("test").getOrCreate()

fields = [StructField("col0", StringType(), True),
          StructField("col1", StringType(), True),
          StructField("col2", StringType(), True),
          StructField("col3", StringType(), True)]
schema = StructType(fields)

chem_info = spark.createDataFrame([], schema)

这是我得到的错误：

  File
"/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/sql/session.py",
line 749, in createDataFrame
    jrdd = self._jvm.SerDeUtil.toJavaArray(rdd._to_java_object_rdd())   File
"/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 2297, in _to_java_object_rdd
    rdd = self._pickled()   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 196, in _pickled
    return self._reserialize(AutoBatchedSerializer(PickleSerializer()))   File
"/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 594, in _reserialize
    self = self.map(lambda x: x, preservesPartitioning=True)   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 325, in map
    return self.mapPartitionsWithIndex(func, preservesPartitioning)   File
"/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 365, in mapPartitionsWithIndex
    return PipelinedRDD(self, f, preservesPartitioning)   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 2514, in __init__
    self.is_barrier = prev._is_barrier() or isFromBarrier   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/rdd.py",
line 2414, in _is_barrier
    return self._jrdd.rdd().isBarrier()   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/py4j/java_gateway.py",
line 1257, in __call__
    answer, self.gateway_client, self.target_id, self.name)   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/pyspark/sql/utils.py",
line 63, in deco
    return f(*a, **kw)   File "/home/.conda/envs/my-env/lib/python3.6/site-packages/py4j/protocol.py",
line 332, in get_return_value
    format(target_id, ".", name, value)) py4j.protocol.Py4JError: An error occurred while calling o57.isBarrier. Trace: py4j.Py4JException:
Method isBarrier([]) does not exist
        at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:318)
        at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:326)
        at py4j.Gateway.invoke(Gateway.java:274)
        at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:132)
        at py4j.commands.CallCommand.execute(CallCommand.java:79)
        at py4j.GatewayConnection.run(GatewayConnection.java:238)
        at java.lang.Thread.run(Thread.java:748)

你能帮我解决吗？

【参考方案1】：

如pyspark: Method isBarrier([]) does not exist 问题中所述，此错误是由安装在 Dataproc 集群中的不同版本的 Spark 与您在 conda 环境中手动安装的 PySpark 之间的不兼容引起的。

要解决此问题，您需要检查集群上的 Spark 版本并安装适当版本的 PySpark：

$ spark-submit --version
Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 2.4.4
      /_/

Using Scala version 2.12.10, OpenJDK 64-Bit Server VM, 1.8.0_232

$ conda install pyspark==2.4.4