一.查找需要清洗的文件
1.1查看hadoopnamenode-log文件位置
1.2 开启Hadoop集群和Hive元数据、Hive远程连接
具体如何开启可以看我之前的文章:(10条消息) SparkSQL-liunx系统Spark连接Hive_难以言喻wyy的博客-CSDN博客
1.3 将这个文件传入到hdfs中:
hdfs dfs -put hadoop-root-namenode-gree2.log /tmp/hadoopNamenodeLogs/hadooplogs/hadoop-root-namenode-gree2.log
二.日志分析
将里面部分字段拿出来分析:
2023-02-10 16:55:33,123 INFO org.apache.hadoop.hdfs.server.namenode.NameNode: registered UNIX signal handlers for [TERM, HUP, INT]
2023-02-10 16:55:33,195 INFO org.apache.hadoop.hdfs.server.namenode.NameNode: createNameNode []
2023-02-10 16:55:33,296 INFO org.apache.hadoop.metrics2.impl.MetricsConfig: loaded properties from hadoop-metrics2.properties
2023-02-10 16:55:33,409 INFO org.apache.hadoop.metrics2.impl.MetricsSystemImpl: Scheduled Metric snapshot period at 10 second(s).
可以看出其可以以INFO来作为中间字段,用indexof读取出该位置索引,以截取字符段的方式来将清洗的数据拿出。
三.代码实现
3.1 对数据进行清洗
object hadoopDemo {def main(args: Array[String]): Unit = {val spark: SparkSession = SparkSession.builder().master("local[*]").appName("HadoopLogsEtlDemo").getOrCreate()val sc: SparkContext = spark.sparkContextimport spark.implicits._import org.apache.spark.sql.functions._
// TODO 根据INFO这个字段来对数据进行封装到Row中。val row: RDD[Row] = sc.textFile("hdfs://192.168.61.146:9000/tmp/hadoopNamenodeLogs/hadooplogs/hadoop-root-namenode-gree2.log").filter(x => {x.startsWith("2023")}).map(x => {val strings: Array[String] = x.split(",")val num1: Int = strings(1).indexOf(" INFO ")val num2: Int = strings(1).indexOf(":")if(num1!=(-1)){val str1: String = strings(1).substring(0, num1)val str2: String = strings(1).substring(num1 + 5, num2)val str3: String = strings(1).substring(num2 + 1, strings(1).length)Row(strings(0), str1, "INFO",str2, str3)}else {val num3: Int = strings(1).indexOf(" WARN ")val num4: Int = strings(1).indexOf(" ERROR ")if(num3!=(-1)&&num4==(-1)){val str1: String = strings(1).substring(0, num3)val str2: String = strings(1).substring(num3 + 5, num2)val str3: String = strings(1).substring(num2 + 1, strings(1).length)Row(strings(0), str1,"WARN", str2, str3)}else{val str1: String = strings(1).substring(0, num4)val str2: String = strings(1).substring(num4 + 6, num2)val str3: String = strings(1).substring(num2 + 1, strings(1).length)Row(strings(0), str1,"ERROR", str2, str3)}}})val schema: StructType = StructType(Array(StructField("event_time", StringType),StructField("number", StringType),StructField("status", StringType),StructField("util", StringType),StructField("info", StringType),))val frame: DataFrame = spark.createDataFrame(row, schema)frame.show(80,false)}}
清洗后的效果图:
3.2 创建jdbcUtils来将其数据导入到数据库:
object jdbcUtils {val url = "jdbc:mysql://192.168.61.141:3306/jsondemo?createDatabaseIfNotExist=true"val driver = "com.mysql.cj.jdbc.Driver"val user = "root"val password = "root"val table_access_logs: String = "access_logs"val table_full_access_logs: String = "full_access_logs"val table_day_active:String="table_day_active"val table_retention:String="retention"val table_loading_json="loading_json"val table_ad_json="ad_json"val table_notification_json="notification_json"val table_active_background_json="active_background_json"val table_comment_json="comment_json"val table_praise_json="praise_json"val table_teacher_json="teacher_json"val properties = new Properties()properties.setProperty("user", jdbcUtils.user)properties.setProperty("password", jdbcUtils.password)properties.setProperty("driver", jdbcUtils.driver)def dataFrameToMysql(df: DataFrame, table: String, op: Int = 1): Unit = {if (op == 0) {df.write.mode(SaveMode.Append).jdbc(jdbcUtils.url, table, properties)} else {df.write.mode(SaveMode.Overwrite).jdbc(jdbcUtils.url, table, properties)}}def getDataFtameByTableName(spark:SparkSession,table:String):DataFrame={val frame: DataFrame = spark.read.jdbc(jdbcUtils.url, table, jdbcUtils.properties)frame}}
3.3 数据导入
jdbcUtils.dataFrameToMysql(frame,jdbcUtils.table_day_active,1)