在交互环境下使用 Pyspark 提交任务给 Spark 解决 : java.sql.SQLException: No suitable driver

在 jupyter 上启用 local 交互环境和 spark 进行交互使用 imapla 来帮助 spark 取数据却失败了

from pyspark.sql import SparkSession

jdbc_url= "jdbc:impala://data1.hundun-new.sa:21050/rawdata;UseNativeQuery=1"
spark = SparkSession.builder 
.appName("sa-test") 
.master("local") 
.getOrCreate()

# properties = {
#     "driver": "com.cloudera.ImpalaJDBC41",
#     "AuthMech": "1",
# #     "KrbRealm": "EXAMPLE.COM",
# #     "KrbHostFQDN": "impala.example.com",
#     "KrbServiceName": "impala"
# }


# df = spark.read.jdbc(url=jdbc_url, table="(/*SA(default)*/ SELECT date, event, count(*) AS c FROM events WHERE date=CURRENT_DATE() GROUP BY 1,2) a")
df = spark.read.jdbc(url=jdbc_url, table="(/*SA(production)*/ SELECT date, event, count(*) AS c FROM events WHERE date=CURRENT_DATE())")
df.select(df['date'], df['event'], df['c'] * 10000).show()


y4JJavaError: An error occurred while calling o32.jdbc.
: java.sql.SQLException: No suitable driver
    at java.sql.DriverManager.getDriver(DriverManager.java:315)
    at org.apache.spark.sql.execution.datasources.jdbc.JDBCOptions$$anonfun$6.apply(JDBCOptions.scala:105)
    at org.apache.spark.sql.execution.datasources.jdbc.JDBCOptions$$anonfun$6.apply(JDBCOptions.scala:105)
    at scala.Option.getOrElse(Option.scala:121)
    at org.apache.spark.sql.execution.datasources.jdbc.JDBCOptions.<init>(JDBCOptions.scala:104)
    at org.apache.spark.sql.execution.datasources.jdbc.JDBCOptions.<init>(JDBCOptions.scala:35)
    at org.apache.spark.sql.execution.datasources.jdbc.JdbcRelationProvider.createRelation(JdbcRelationProvider.s

可以清楚的看到报出的错误 No suitable driver ，我们需要添加上 impala 的 jdbc driver 才能正常运行。

首先我们下载一个 impala 的 jdbc driver 

http://repo.odysseusinc.com/artifactory/community-libs-release-local/com/cloudera/ImpalaJDBC41/2.6.3/ImpalaJDBC41-2.6.3.jar

然后我们在申请 ss 的时候通过 cnofig 指定该 impala driver 的路径即可

from pyspark.sql import SparkSession

jdbc_url= "jdbc:impala://data1.hundun-new.sa:21050/rawdata;UseNativeQuery=1"
spark = SparkSession.builder 
.appName("sa-test") 
.master("local") 
.config('spark.driver.extraClassPath', '/usr/share/java/ImpalaJDBC41-2.6.3.jar') 
.getOrCreate()

这里我在 stackoverflow 上还找到另外一种方法

EDIT

The answers from How to load jar dependenices in IPython Notebook are already listed in the link I shared myself, and do not work for me. I already tried to configure the environment variable from the notebook:

import os
os.environ['PYSPARK_SUBMIT_ARGS'] = '--driver-class-path /path/to/postgresql.jar --jars /path/to/postgresql.jar'

There's nothing wrong with the file path or the file itself since it works fine when I specify it and run the pyspark-shell.

Reference:

https://spark.apache.org/docs/latest/configuration.html    Spark Configuration 

https://stackoverflow.com/questions/51772350/how-to-specify-driver-class-path-when-using-pyspark-within-a-jupyter-notebook    How to specify driver class path when using pyspark within a jupyter notebook?