How to join 2 sql spark streams

ENV: Scala spark version: 2.1.1

These are my streams (read from kafka):

val conf = new SparkConf()
  .setMaster("local[1]")
  .setAppName("JoinStreams")

val spark = SparkSession.builder().config(conf).getOrCreate()

import spark.implicits._

val schema = StructType(
  List(
    StructField("t", DataTypes.StringType),
    StructField("dst", DataTypes.StringType),
    StructField("dstPort", DataTypes.IntegerType),
    StructField("src", DataTypes.StringType),
    StructField("srcPort", DataTypes.IntegerType),
    StructField("ts", DataTypes.LongType),
    StructField("len", DataTypes.IntegerType),
    StructField("cpu", DataTypes.DoubleType),
    StructField("l", DataTypes.StringType),
    StructField("headers", DataTypes.createArrayType(DataTypes.StringType))
  )
)
val baseDataFrame = spark
  .readStream
  .format("kafka")
  .option("kafka.bootstrap.servers", "host:port")
  .option("subscribe", 'topic')
  .load()
  .selectExpr("cast (value as string) as json")
  .select(from_json($"json", schema).as("data"))
  .select($"data.*")

val requestsDataFrame = baseDataFrame
  .filter("t = 'REQUEST'")
  .repartition($"dst")
  .withColumn("rowId", monotonically_increasing_id())

val responseDataFrame = baseDataFrame
  .filter("t = 'RESPONSE'")
  .repartition($"src")
  .withColumn("rowId", monotonically_increasing_id())

responseDataFrame.createOrReplaceTempView("responses")
requestsDataFrame.createOrReplaceTempView("requests")


val dataFrame = spark.sql("select * from requests left join responses ON requests.rowId = responses.rowId")

      

        
        
        
      

    

I am getting this ERROR when starting the application:

org.apache.spark.sql.AnalysisException: Left outer/semi/anti joins with a streaming DataFrame/Dataset on the right is not supported;;

      

        
        
        
      

    

How can I join these two threads? I am also trying to make a direct connection and get the same error. Should I first save it to a file and then read it again? What's the best practice?