Pyspark 中的 Kafka “partition.assignment.strategy”

Question

我正在尝试读取数据以将其转换为 Dataframe 并且我的软件的当前版本如下：

1. spark-2.4.7-bin-hadoop2.7
2. kafka_2.12-2.7.0

卡夫卡正在工作，我存储了以下数据，我正在尝试读取：

~/development/kafka_home/kafka_2.13-2.6.0$ bin/kafka-console-consumer.sh --bootstrap-server localhost:9092 --topic testtopic --from-beginning
{"transaction_id": "1", "transaction_card_type": "Visa", "transaction_amount": 181.76, "transaction_datetime": "2021-01-25 15:44:44"}
{"transaction_id": "2", "transaction_card_type": "MasterCard", "transaction_amount": 228.62, "transaction_datetime": "2021-01-25 15:44:45"}
{"transaction_id": "3", "transaction_card_type": "Visa", "transaction_amount": 483.48, "transaction_datetime": "2021-01-25 15:44:46"}
{"transaction_id": "4", "transaction_card_type": "MasterCard", "transaction_amount": 477.87, "transaction_datetime": "2021-01-25 15:44:47"}
{"transaction_id": "5", "transaction_card_type": "MasterCard", "transaction_amount": 304.52, "transaction_datetime": "2021-01-25 15:44:48"}
{"transaction_id": "1", "transaction_card_type": "MasterCard", "transaction_amount": 346.99, "transaction_datetime": "2021-01-25 16:38:44"}
{"transaction_id": "2", "transaction_card_type": "Maestro", "transaction_amount": 384.33, "transaction_datetime": "2021-01-25 16:38:45"}
{"transaction_id": "3", "transaction_card_type": "MasterCard", "transaction_amount": 394.95, "transaction_datetime": "2021-01-25 16:38:46"}
{"transaction_id": "4", "transaction_card_type": "Visa", "transaction_amount": 22.75, "transaction_datetime": "2021-01-25 16:38:47"}
{"transaction_id": "5", "transaction_card_type": "MasterCard", "transaction_amount": 492.01, "transaction_datetime": "2021-01-25 16:38:48"}

我在 PySpark 中执行以下代码

from pyspark.sql import SparkSession
from pyspark.sql.functions import *
from pyspark.sql.types import *

KAFKA_TOPIC_NAME_CONS = "testtopic"
KAFKA_BOOTSTRAP_SERVERS_CONS = 'localhost:9092'

spark = SparkSession \
    .builder \
    .appName("PySpark Structured Streaming with Kafka Demo") \
    .config("spark.jars", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/kafka-clients-1.1.0.jar") \
    .config("spark.jars", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-8-assembly_2.11-2.4.7.jar") \
    .config("spark.jars", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-sql-kafka-0-10_2.11-2.4.7.jar") \
    .config("spark.executor.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/kafka-clients-1.1.0.jar") \
    .config("spark.executor.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-8-assembly_2.11-2.4.7.jar") \
    .config("spark.executor.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-sql-kafka-0-10_2.11-2.4.7.jar") \
    .config("spark.driver.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/kafka-clients-1.1.0.jar") \
    .config("spark.driver.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-8-assembly_2.11-2.4.7.jar") \
    .config("spark.driver.extraClassPath", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-sql-kafka-0-10_2.11-2.4.7.jar") \
    .config("spark.executor.extraLibrary", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/kafka-clients-1.1.0.jar") \
    .config("spark.executor.extraLibrary", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-8-assembly_2.11-2.4.7.jar") \
    .config("spark.executor.extraLibrary", "/home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-sql-kafka-0-10_2.11-2.4.7.jar") \
    .getOrCreate()

df = spark.readStream.format("kafka").option("kafka.bootstrap.servers", "localhost:9092").option("subscribe", "testtopic").load()
ds = df.selectExpr("CAST(key AS STRING)", "CAST(value AS STRING)")
query = ds \
  .writeStream \
  .queryName("tableName") \
  .option("partition.assignment.strategy", "range")
  .format("console") \
  .start()

我得到的错误如下：

21/01/25 18:53:41 WARN kafka010.KafkaOffsetReader：尝试 1 获取 Kafka 偏移量时出错：org.apache.kafka.common.config.ConfigException：缺少所需的配置“partition.assignment.strategy”，没有默认值.

我做了一些研究，他们说名为“kafka-clients-1.1.0.jar”的.jar 文件似乎是问题所在，但是我已经训练了 2.6.0 和 1.1.0 版本，结果相同。

**

编辑：

**

我在“spark-defaults”中添加了以下内容

spark.jars /home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-10_2.12-2.4.7.jar
spark.executor.extraClassPath /home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-10_2.12-2.4.7.jar
spark.driver.extraClassPath /home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-10_2.12-2.4.7.jar
spark.executor.extraLibrary /home/bupry_dev/development/spark_home/spark-2.4.7-bin-hadoop2.7/jars/spark-streaming-kafka-0-10_2.12-2.4.7.jar

并通过以下方式创建我的 Session：

spark = SparkSession \
    .builder \
    .appName("PySpark Structured Streaming with Kafka Demo") \
    .getOrCreate()

我仍然收到以下错误：

java.util.ServiceConfigurationError: org.apache.spark.sql.sources.DataSourceRegister: Provider org.apache.spark.sql.kafka010.KafkaSourceProvider could not be instantiated

对于这行代码：

df = spark.readStream.format("kafka").option("kafka.bootstrap.servers", "localhost:9092").option("subscribe", "testtopic").load()

Answer 1

如Spark docs中所述，您只需要包含以下依赖项：

groupId = org.apache.spark
artifactId = spark-streaming-kafka-0-10_2.11
version = 2.4.7 <-- replace this by your appropriate Spark version

Spark 警告不要直接使用kafka-clients*.jar ，因为它已经包含了那些 jars，并且为同一个库添加多个 jars 会使调试更加困难。

不要手动添加对 org.apache.kafka 工件（例如 kafka-clients）的依赖。 spark-streaming-kafka-0-10 工件已经具有适当的传递依赖关系，并且不同的版本可能以难以诊断的方式不兼容。