时间戳

首页 > 解决方案 > 无法从 spark 读取 Azure Eventhub 主题

apache-spark - 无法从 spark 读取 Azure Eventhub 主题

问题描述

环境细节

  1. 火花版本：3.x
  2. Python 3.8 版和 Java 8 版
  3. 天蓝色-eventhubs-spark_2.12-2.3.17.jar
import json
from pyspark.sql import SparkSession


#the below command getOrCreate() uses the SparkSession shared across the jobs instead of using one SparkSession per job.
spark = SparkSession.builder.appName('ntorq_eventhub_load').getOrCreate()

#ntorq adls checkpoint location.
ntorq_connection_string = "connection-string"

ehConf = {}
ehConf['eventhubs.connectionString'] = spark.sparkContext._jvm.org.apache.spark.eventhubs.EventHubsUtils.encrypt(ntorq_connection_string)
# ehConf['eventhubs.connectionString'] = ntorq_connection_string
ehConf['eventhubs.consumerGroup'] = "$default"

OFFSET_START = "-1"   # the beginning
OFFSET_END = "@latest"

# Create the positions
startingEventPosition = {
  "offset": OFFSET_START ,
  "seqNo": -1,            #not in use
  "enqueuedTime": None,   #not in use
  "isInclusive": True
}

endingEventPosition = {
  "offset": OFFSET_END,           #not in use
  "seqNo": -1,              #not in use
  "enqueuedTime": None,
  "isInclusive": True
}

# Put the positions into the Event Hub config dictionary
ehConf["eventhubs.startingPosition"] = json.dumps(startingEventPosition)
ehConf["eventhubs.endingPosition"] = json.dumps(endingEventPosition)


df = spark \
  .readStream \
  .format("eventhubs") \
  .options(**ehConf) \
  .load() \
  .selectExpr("cast(body as string) as body_str")

df.writeStream \
    .format("console") \
    .start()

错误

21/04/25 20:17:53 WARN Utils: Your hostname,resolves to a loopback address: 127.0.0.1; using 192.168.1.202 instead (on interface en0)
21/04/25 20:17:53 WARN Utils: Set SPARK_LOCAL_IP if you need to bind to another address
21/04/25 20:17:53 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Using Spark's default log4j profile: org/apache/spark/log4j-defaults.properties
Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
Traceback (most recent call last):
  File "/Users/PycharmProjects/pythonProject/test.py", line 12, in <module>
    ehConf['eventhubs.connectionString'] = spark.sparkContext._jvm.org.apache.spark.eventhubs.EventHubsUtils.encrypt(ntorq_connection_string)
TypeError: 'JavaPackage' object is not callable

代码在 databricks 环境中运行良好但无法使用来自 eventthub 的所有消息我尝试在每次运行之前清除默认检查点文件夹但仍然面临问题，所以想在本地系统上尝试。在尝试面临 JavaPackage 问题的本地环境时。感谢任何帮助。谢谢你

标签： apache-sparkpysparkazure-databricks

解决方案

创建会话时需要添加 EventHubs 包：

park = SparkSession.builder.appName('ntorq_eventhub_load')\
  .config("spark.jars.packages", "com.microsoft.azure:azure-eventhubs-spark_2.12:2.3.18")\
  .getOrCreate()

推荐阅读