spark hbasefilter hbase

  1. package cn.deppon.Tool  
  2.   
  3. import java.util  
  4.   
  5. import scala.collection.JavaConverters._  
  6. import org.apache.hadoop.hbase.HBaseConfiguration  
  7. import org.apache.hadoop.hbase.client.{HTable, ResultScanner, Scan}  
  8. import org.apache.hadoop.hbase.filter._  
  9. import org.apache.hadoop.hbase.mapreduce.TableInputFormat  
  10. import org.apache.hadoop.hbase.protobuf.ProtobufUtil  
  11. import org.apache.hadoop.hbase.util.{Base64, Bytes}  
  12. import org.apache.log4j.{Level, Logger}  
  13. import org.apache.spark.sql.{Row, SQLContext}  
  14. import org.apache.spark.sql.types._  
  15. import org.apache.spark.{SparkConf, SparkContext}  
  16.   
  17. /** 
  18.   * Created by DreamBoy on 2017/5/12. 
  19.   */  
  20. object SparkHbaseTool {  
  21.   
  22.   /** 
  23.     * 利用主构造器构造需要的环境的基本条件 
  24.     */  
  25.   Logger.getLogger("org.apache.spark").setLevel(Level.ERROR)  
  26.   //设置spark参数  
  27.   val conf = new SparkConf().setMaster("local[2]")  
  28.     .setAppName("HbaseTest")  
  29.   conf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer")  
  30.   val sc = new SparkContext(conf)  
  31.   val hbaseConf = HBaseConfiguration.create()  
  32.   val sqlContext = new SQLContext(sc)  
  33.   
  34.   //配置HBase  
  35.   hbaseConf.set("hbase.rootdir", "hdfs://http://192.168.10.228/hbase")  
  36.   hbaseConf.set("hbase.zookeeper.quorum", "192.168.10.228,192.168.10.229,192.168.10.230,192.168.10.231,192.168.10.232")  
  37.   hbaseConf.set("hbase.zookeeper.property.clientPort", "2181")  
  38.   hbaseConf.set("hbase.master", "192.168.10.230")  
  39.   
  40.   def convertScanToString(scan: Scan) = {  
  41.     val proto = ProtobufUtil.toScan(scan)  
  42.     Base64.encodeBytes(proto.toByteArray)  
  43.   }  
  44.   
  45.   /** 
  46.     * 
  47.     * @param tbl_nm     表名 
  48.     * @param show_col   _1 列族  _2列名  _3 列类型(String,Int,Double,Timestamp...) 
  49.     * @param filter_col _1 列族  _2列名  _3 筛选值  _4 筛选类型(=,<,>,!=...) 
  50.     * @return sqlcontext 
  51.     */  
  52.   def getTableNm(tbl_nm: String, show_col: Array[(String, String, String)], filter_col: Array[(String, String, String, String)]): (SQLContext) = {  
  53.   
  54.     hbaseConf.set(TableInputFormat.INPUT_TABLE, tbl_nm)  
  55.     val table = new HTable(hbaseConf, tbl_nm)  
  56.     val scan = new Scan()  
  57.   
  58.     /** 
  59.       * 指定列族和需要显示的列名 
  60.       * 添加多个需要用到的列 
  61.       */  
  62.     /* 
  63.     val length = show_col.length 
  64.     for(i <- show_col){ 
  65.       scan.addColumn(Bytes.toBytes(i._1),Bytes.toBytes(i._2)) 
  66.     } 
  67.     */  
  68.     //设置rowkey的范围，启示和结束  
  69.     //scan.setStartRow(Bytes.toBytes(""))  
  70.     //scan.setStopRow(Bytes.toBytes(""))  
  71.     val fil_len = filter_col.length  
  72.     println("------>>>>" + fil_len)  
  73.     //如果没有添加过滤器，就给过滤器添加空  
  74.     if (fil_len > 0) {  
  75.       val filter_arr = new util.ArrayList[Filter](fil_len)  
  76.   
  77.       for (i <- filter_col) {  
  78.         i._4 match {  
  79.           case "=" => {  
  80.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  81.               Bytes.toBytes(i._2), CompareFilter.CompareOp.EQUAL, new BinaryComparator(Bytes.toBytes(i._3)))  
  82.             filter1.setFilterIfMissing(true)  
  83.             filter_arr.add(filter1)  
  84.           }  
  85.           case "<" => {  
  86.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  87.               Bytes.toBytes(i._2), CompareFilter.CompareOp.LESS, new BinaryComparator(Bytes.toBytes(i._3)))  
  88.             filter1.setFilterIfMissing(true)  
  89.             filter_arr.add(filter1)  
  90.           }  
  91.           case "<=" => {  
  92.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  93.               Bytes.toBytes(i._2), CompareFilter.CompareOp.LESS_OR_EQUAL, new BinaryComparator(Bytes.toBytes(i._3)))  
  94.             filter1.setFilterIfMissing(true)  
  95.             filter_arr.add(filter1)  
  96.           }  
  97.           case ">" => {  
  98.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  99.               Bytes.toBytes(i._2), CompareFilter.CompareOp.GREATER, new BinaryComparator(Bytes.toBytes(i._3)))  
  100.             filter1.setFilterIfMissing(true)  
  101.             filter_arr.add(filter1)  
  102.           }  
  103.           case ">=" => {  
  104.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  105.               Bytes.toBytes(i._2), CompareFilter.CompareOp.GREATER_OR_EQUAL, new BinaryComparator(Bytes.toBytes(i._3)))  
  106.             //filter1.setFilterIfMissing(true)  
  107.             filter_arr.add(filter1)  
  108.           }  
  109.           case "!=" => {  
  110.             val filter1 = new SingleColumnValueFilter(Bytes.toBytes(i._1),  
  111.               Bytes.toBytes(i._2), CompareFilter.CompareOp.NOT_EQUAL, new BinaryComparator(Bytes.toBytes(i._3)))  
  112.             filter1.setFilterIfMissing(true)  
  113.             filter_arr.add(filter1)  
  114.           }  
  115.           case _ => {}  
  116.         }  
  117.       }  
  118.       /** 
  119.         * 通过使用filterlist可以加载多个过滤器 
  120.         * 设置多个过滤器 
  121.         */  
  122.       val filterList = new FilterList(FilterList.Operator.MUST_PASS_ALL, filter_arr)  
  123.       scan.setFilter(filterList)  
  124.     } else {  
  125.       scan.setFilter(null)  
  126.     }  
  127.   
  128.     //hbaseConf.set(TableInputFormat.SCAN,convertScanToString(scan))  
  129.     //获取表的扫描  
  130.     val ColumnValueScanner = table.getScanner(scan)  
  131.     //构建structtype需要的list  根据传入的类型参数构建表  
  132.     /*var list_col = show_col.map{x=>{ 
  133.      /* x._3 match { 
  134.         case "String" => StructField(x._2,StringType,true) 
  135.         case "Int" => StructField(x._2,StringType,true) 
  136.         case "Double" => StructField(x._2,StringType,true) 
  137.         case "Timestamp" => StructField(x._2,StringType,true) 
  138.         case _ => StructField(x._2,StringType,true) 
  139.       }*/  
  140.       StructField(x._2,StringType,true)  
  141.     }  
  142.     }*/  
  143.     /** 
  144.       * structType构造的目的：为在后面产生dataframe的时候指定每个值的列名 
  145.       * 在注册成表的时候可以使用 
  146.       */  
  147.     var list_col: List[StructField] = List()  
  148.     list_col :+= StructField("id", StringType, true)  
  149.       
  150.     for (i <- show_col) {  
  151.       list_col :+= StructField(i._2, StringType, true)  
  152.     }  
  153.   
  154.       
  155.     //构建表的structType  
  156.     val schema = StructType(list_col)  
  157.   
  158.     val tbl_rdd = ColumnValueScanner.iterator().asScala  
  159.     //把过滤器加载到hbaseconf中  
  160.     hbaseConf.set(TableInputFormat.SCAN, convertScanToString(scan))  
  161.     //构建RDD  
  162.     val hbaseRDD = sc.newAPIHadoopRDD(  
  163.       hbaseConf,  
  164.       classOf[TableInputFormat],  
  165.       classOf[org.apache.hadoop.hbase.io.ImmutableBytesWritable],  
  166.       classOf[org.apache.hadoop.hbase.client.Result])  
  167.     //构建rdd的结果集  
  168.     val rowRDD = hbaseRDD.map { case (_, result) => {  
  169.       var valueSeq: Seq[String] = Seq()  
  170.       //获取行键  
  171.       val key = Bytes.toString(result.getRow)  
  172.   
  173.       //通过列族和列名获取列  不加rowkey方法  
  174.       //      for(column <- columns) {  
  175.       //        valueSeq :+= Bytes.toString(result.getValue(family.getBytes, column.getBytes))  
  176.       //      }  
  177.       //加rowkey方法，Array第一列必须是"rowkey"  
  178.       valueSeq :+= key  
  179.       for (row <- show_col) {  
  180.         valueSeq :+= Bytes.toString(result.getValue(row._1.getBytes, row._2.getBytes))  
  181.       }  
  182.       Row.fromSeq(valueSeq)  
  183.     }  
  184.     }  
  185.     val hbasedataframe = sqlContext.createDataFrame(rowRDD, schema)  
  186.   
  187.     hbasedataframe.registerTempTable(tbl_nm)  
  188.     sqlContext  
  189.   }  
  190. }

Commentaires

Enregistrer un commentaire

Posts les plus consultés de ce blog

Controlling Parallelism in Spark by controlling the input partitions by controlling the input partitions

Image
It is useful to be able to control the degree of parallelism when using Spark. Spark provides a very convenient method to increase the degree of parallelism which should be adequate in practice. This blog entry of the "Inside Spark" series describes the knobs Spark uses to control the degree of parallelism. Controlling the number of partitions in the " local " mode Spark is designed ground up to enable unit testing. To control the degree of parallelism in the local mode simply utilize the function JavaRDD<String> org.apache.spark.api.java.JavaSparkContext.parallelize(List<String> list, int numSlices) The following snippet of code demonstrates this usage- public class SparkControlPartitionSizeLocally { public static void main ( String [] args ) { JavaSparkContext sc = new JavaSparkContext ( "local" , "localpartitionsizecontrol" ); String input = "four score and sev
Lire la suite

Spark performance optimization: shuffle tuning

Image
Article directory 1 shuffle tuning 1.1 Summary of tuning 1.2 Overview of ShuffleManager Development 1.3 HashShuffleManager operating principle 1.3.1 Unoptured HashShuffleManager 1.3.2 Optimized HashShuffleManager 1.4 SortShuffleManager operating principle 1.4.1 General operating mechanism 1.4.2 bypass running mechanism 1.5 shuffle related parameters tuning 1.5.1 spark.shuffle.file.buffer 1.5.2 spark.reducer.maxSizeInFlight 1.5.3 spark.shuffle.io.maxRetries 1.5.4 spark.shuffle.io.retryWait 1.5.5 spark.shuffle.memoryFraction 1.5.6 spark.shuffle.manager 1.5.7 spark.shuffle.sort.bypassMergeThreshold 1.5.8 spark.shuffle.consolidateFiles 2 write in the last words Shuffle Summary of tuning Most of the performance of Spark operations is mainly consumed in the shuffle link, because the link contains a large number of disk IO, serialization, network data transmission and othe
Lire la suite

Spark optimization

Image
One operation and maintenance 1. Master hang up, standby restart is also invalid Master defaults to 512M of memory, when the task in the cluster is particularly high, it will hang, because the master will read each task event log log to generate spark ui, the memory will naturally OOM, you can run the log See that the master of the start through the HA will naturally fail for this reason. solve Increase the Master's memory spark-env.sh , set in the master node spark-env.sh : export SPARK_DAEMON_MEMORY 10g # 根据你的实际情况 Reduce the job information stored in the Master memory spark.ui.retainedJobs 500 # 默认都是1000 spark.ui.retainedStages 500 Hang up or suspend Sometimes we will see the web node in the web ui disappear or in the dead state, the task of running the node will report a variety of lost worker errors, causing the same reasons and the above, worker memory to save a lot of ui The information leads to gc when the heartbeat is lost
Lire la suite