本地spark测试tomcat日志处理_tomcat spark-优快云博客

tomcat访问日志分隔符用tab（利用了tab的unicode表示，“	”），因为不容易出现在各种日志变量的内容中：

tomcat日志的格式化（用表示tab替换默认的空格分隔符）：

<Valve className="org.apache.catalina.valves.AccessLogValve" directory="logs"

prefix="localhost_access_log" suffix=".txt"

pattern="%h%l %u %t "%r" %s %b %D" />

日志的封装处理

package sparkcore

case class LogField(
                     ip: String,
                     time: String,
                     request: String,
                     status: Int,
                     length: Int)

object LogField {
  def parseLog(log: String): LogField = {
    val fields = log.split("\t");
    val len = if (fields(6) == "-") 0 else fields(6).toLong //注意响应字节为0的时候，日志记录为“-”
    LogField(fields(0), fields(3), fields(4), fields(5).toInt, len);
  }

  def main(args: Array[String]): Unit = {
    val log = "192.68.1.134\t-\t-\t[05/Oct/2017:17:36:41 +0800]\t\"GET /wget/finished/ HTTP/1.1\"\t200\t-";
    val fields = parseLog(log)
    println(fields.ip + "\t" + fields.time + "\t" + fields.request + "\t" + fields.status + "\t" + fields.length)
  }
}

注意日志字段的处理，特别是需要类型转换的。

package sparkcore

import org.apache.spark.{SparkConf, SparkContext}

object LogAnalyer {
  def main(args: Array[String]): Unit = {
    val conf = new SparkConf().setMaster("local").setAppName("LogAnalyer")
    val sc = new SparkContext(conf);
    val logsRDD = sc.textFile("file:///D:/data/log.txt")
      .map(line => LogField.parseLog(line))
      .cache()
    /**
      *统计web服务器所有响应中的最大、最小及平均字节数
      */
    val contextSize = logsRDD.map { log => log.length }
    val maxSize = contextSize.max()
    val minSize = contextSize.min()
    val averageSize = contextSize.reduce(_ + _) / contextSize.count()
    println("响应最大值：" + maxSize + "  最小值：" + minSize + "   平均值：" + averageSize);

    /**
      * 统计各种响应状态的出现次数
      */
    logsRDD.map(log => (log.status, 1))
      .reduceByKey(_ + _)
      .foreach(result => println(" 响应状态：" + result._1 + "  计数：" + result._2))

    /**
      * 统计访问总次数超过1000的前3名的ip
      */
    val result = logsRDD.map(log => (log.ip, 1))
      .reduceByKey(_ + _)
      .filter(result => result._2 > 1000)
      .map(m => (m._2, m._1))
      .top(3)
    for (tuple <- result) {
      println("ip : " + tuple._2 + "  请求次数：" + tuple._1);
    }

    /**
      * 统计请求URI的TopN
      */
    val topN = logsRDD.map { log => (log.request.split(" ")(1), 1) }
      .reduceByKey(_ + _)
      .map(result => (result._2, result._1))
      .sortByKey(false)
      .take(3)
    for (tuple <- topN) {
      println("URI : " + tuple._2 + "  请求频次：" + tuple._1);
    }

    //最后一定要释放内存
    logsRDD.unpersist(true)
  }
}