spark读取sqlserver数据BulkLoad到hbase

def load2hbase2(sc: SparkContext,tableName: String): Unit ={
    val sqlContext = new SQLContext(sc)
    val properties = new Properties()

    properties.put("user","xxx")
    properties.put("password","xxx")
    properties.put("driver","com.microsoft.sqlserver.jdbc.SQLServerDriver")
    val url = "jdbc:sqlserver://xxxx:1433;databaseName=xxxx"

    //分区条件,sID是uuid字符串，取第一个字符就是a-f和0-9分16个区
    val arr = new ArrayBuffer[String]()
    arr += "substring(sID,1,1)='a'"
    arr += "substring(sID,1,1)='d'"
    arr += "substring(sID,1,1)='f'"
    arr += "substring(sID,1,1)='6'"
    arr += "substring(sID,1,1)='c'"
    arr += "substring(sID,1,1)='8'"
    arr += "substring(sID,1,1)='2'"
    arr += "substring(sID,1,1)='e'"
    arr += "substring(sID,1,1)='3'"
    arr += "substring(sID,1,1)='4'"
    arr += "substring(sID,1,1)='0'"
    arr += "substring(sID,1,1)='5'"
    arr += "substring(sID,1,1)='7'"
    arr += "substring(sID,1,1)='1'"
    arr += "substring(sID,1,1)='9'"
    arr += "substring(sID,1,1)='b'"

    val conf = HBaseConfiguration.create()

    val table = new HTable(conf, tableName)

    conf.set(TableOutputFormat.OUTPUT_TABLE, tableName)
    conf.set("hbase.zookeeper.quorum","xxx")
    conf.set("hbase.zookeeper.property.clientPort", "2181")
    val job = Job.getInstance(conf)
    job.setMapOutputKeyClass (classOf[ImmutableBytesWritable])
    job.setMapOutputValueClass (classOf[KeyValue])
    HFileOutputFormat.configureIncrementalLoad (job, table)

    val sdf = sqlContext.read.jdbc(url,"(select sID,sCID,sCName,sBID_BL,sBID_Sd from table) t",arr.toArray,properties)
    val rdd = sdf.flatMap(x=>{
      var arr = new ArrayBuffer[(ImmutableBytesWritable,KeyValue)](4)
      if(x.length >= 5 && Prepare.citys.contains(getValue(x.getString(2)))){
        val rkbytes = Bytes.toBytes(getValue(x.getString(0)))
        val rkim = new ImmutableBytesWritable(rkbytes)
        arr += ((rkim,new KeyValue(rkbytes, "all".getBytes(), "SbidBl".getBytes(), Bytes.toBytes(getValue(x.getString(3))))))
        arr += ((rkim,new KeyValue(rkbytes, "all".getBytes(), "SbidSd".getBytes(), Bytes.toBytes(getValue(x.getString(4))))))
        arr += ((rkim,new KeyValue(rkbytes, "all".getBytes(), "Scid".getBytes(), Bytes.toBytes(getValue(x.getString(1))))))
        arr += ((rkim,new KeyValue(rkbytes, "all".getBytes(), "Scname".getBytes(), Bytes.toBytes(getValue(x.getString(2))))))
      }
      arr
    }).filter(x=>x._1!=null).sortBy(_._1)
    val p = "/tmp/test_loadhbase/"+System.currentTimeMillis()

    job.getConfiguration.set("mapred.output.dir", p)
    rdd.saveAsNewAPIHadoopDataset(job.getConfiguration)

    val bulkLoader = new LoadIncrementalHFiles(job.getConfiguration)
    bulkLoader.doBulkLoad(new Path(p), table)
  }