PySpark Example Code

Development Description

The CloudTable OpenTSDB and MRS OpenTSDB can be connected to DLI as data sources.

• Prerequisites

  A datasource connection has been created on the DLI management console.

  NOTE:

  Hard-coded or plaintext passwords pose significant security risks. To ensure security, encrypt your passwords, store them in configuration files or environment variables, and decrypt them when needed.

• Code implementation
  1. Import dependency packages.

     1 2 3

     from __future__ import print_function from pyspark.sql.types import StructType, StructField, StringType, LongType, DoubleType from pyspark.sql import SparkSession

  2. Create a session.

     1	sparkSession = SparkSession.builder.appName("datasource-opentsdb").getOrCreate()

  3. Create a table to connect to an OpenTSDB data source.

     1 2 3 4

     sparkSession.sql("create table opentsdb_test using opentsdb options( 'Host'='opentsdb-3xcl8dir15m58z3.cloudtable.com:4242', 'metric'='ct_opentsdb', 'tags'='city,location')")

     NOTE:

     For details about the Host, metric, and tags parameters, see Table 1.

• Connecting to data sources through SQL APIs
  1. Insert data.

     sparkSession.sql("insert into opentsdb_test values('aaa', 'abc', '2021-06-30 18:00:00', 30.0)")
  2. Query data.

     result = sparkSession.sql("SELECT * FROM opentsdb_test")
• Connecting to data sources through DataFrame APIs
  1. Construct a schema.

     1 2 3 4

     schema = StructType([StructField("location", StringType()),\ StructField("name", StringType()), \ StructField("timestamp", LongType()),\ StructField("value", DoubleType())])

  2. Set data.

     1	dataList = sparkSession.sparkContext.parallelize([("aaa", "abc", 123456L, 30.0)])

  3. Create a DataFrame.

     1	dataFrame = sparkSession.createDataFrame(dataList, schema)

  4. Import data to OpenTSDB.

     1	dataFrame.write.insertInto("opentsdb_test")

  5. Read data from OpenTSDB.

     1 2 3 4 5 6 7

     jdbdDF = sparkSession.read .format("opentsdb")\ .option("Host","opentsdb-3xcl8dir15m58z3.cloudtable.com:4242")\ .option("metric","ctopentsdb")\ .option("tags","city,location")\ .load() jdbdDF.show()

  6. View the operation result.

     

• Submitting a Spark job
  1. Upload the Python code file to DLI.

  2. In the Spark job editor, select the corresponding dependency module and execute the Spark job.
     NOTE:

     • If the Spark version is 2.3.2 (will be offline soon) or 2.4.5, specify the Module to sys.datasource.opentsdb when you submit a job.
     • If the Spark version is 3.1.1, you do not need to select a module. Configure Spark parameters (--conf).

       spark.driver.extraClassPath=/usr/share/extension/dli/spark-jar/datasource/opentsdb/*

       spark.executor.extraClassPath=/usr/share/extension/dli/spark-jar/datasource/opentsdb/*

     • For details about how to submit a job on the console, see the description of the table "Parameters for selecting dependency resources" in Creating a Spark Job.
     • For details about how to submit a job through an API, see the description of the modules parameter in Table 2 "Request parameters" in Creating a Batch Processing Job.

Complete Example Code

• Connecting to MRS OpenTSDB through SQL APIs

  # _*_ coding: utf-8 _*_
  from __future__ import print_function
  from pyspark.sql.types import StructType, StructField, StringType, LongType, DoubleType
  from pyspark.sql import SparkSession
  
  if __name__ == "__main__":
    # Create a SparkSession session.    
    sparkSession = SparkSession.builder.appName("datasource-opentsdb").getOrCreate()
  
  
    # Create a DLI cross-source association opentsdb data table    
    sparkSession.sql(\
      "create table opentsdb_test using opentsdb options(\
      'Host'='10.0.0.171:4242',\
      'metric'='cts_opentsdb',\
      'tags'='city,location')")
  
    sparkSession.sql("insert into opentsdb_test values('aaa', 'abc', '2021-06-30 18:00:00', 30.0)")
  
    result = sparkSession.sql("SELECT * FROM opentsdb_test")
    result.show()
  
    # close session 
    sparkSession.stop()
• Connecting to OpenTSDB through DataFrame APIs

  # _*_ coding: utf-8 _*_
  from __future__ import print_function
  from pyspark.sql.types import StructType, StructField, StringType, LongType, DoubleType
  from pyspark.sql import SparkSession
  
  if __name__ == "__main__":
    # Create a SparkSession session.    
    sparkSession = SparkSession.builder.appName("datasource-opentsdb").getOrCreate()
  
    # Create a DLI cross-source association opentsdb data table    
    sparkSession.sql(
      "create table opentsdb_test using opentsdb options(\
      'Host'='opentsdb-3xcl8dir15m58z3.cloudtable.com:4242',\
      'metric'='ct_opentsdb',\
      'tags'='city,location')")
  
    # Create a DataFrame and initialize the DataFrame data.    
    dataList = sparkSession.sparkContext.parallelize([("aaa", "abc", 123456L, 30.0)])
  
    # Setting schema   
    schema = StructType([StructField("location", StringType()),\                 
                         StructField("name", StringType()),\                
                         StructField("timestamp", LongType()),\               
                         StructField("value", DoubleType())])
  
    # Create a DataFrame from RDD and schema   
    dataFrame = sparkSession.createDataFrame(dataList, schema)
  
    # Set cross-source connection parameters   
    metric = "ctopentsdb"   
    tags = "city,location"  
    Host = "opentsdb-3xcl8dir15m58z3.cloudtable.com:4242"
  
    # Write data to the cloudtable-opentsdb   
    dataFrame.write.insertInto("opentsdb_test")
    # ******* Opentsdb does not currently implement the ctas method to save data, so the save() method cannot be used.*******   
    # dataFrame.write.format("opentsdb").option("Host", Host).option("metric", metric).option("tags", tags).mode("Overwrite").save()   
  
    # Read data on CloudTable-OpenTSDB  
    jdbdDF = sparkSession.read\       
        .format("opentsdb")\      
        .option("Host",Host)\     
        .option("metric",metric)\  
        .option("tags",tags)\    
        .load()   
    jdbdDF.show()
  
    # close session 
    sparkSession.stop()