黑猴子的家：Flume 1.6 移植 1.7 Taildir Source 功能

1、官网地址

http://flume.apache.org/

2、下载Flume1.6 和 Flume1.7

3、下载历史版本

4、历史版本

5、Flume 1.7

6、Flume 1.6

7、解压Flume1.7 找到下面6个类

PollableSourceConstants.java
ReliableTaildirEventReader.java
TaildirMatcher.java
TaildirSource.java
TaildirSourceConfigurationConstants.java
TailFile.java

8、ReliableTaildirEventReader.java 文件updateTailFiles 方法 源码修改

修改处，标有TODO，有两处修改

/**
   * Update tailFiles mapping if a new file is created or appends are detected
   * to the existing file.
   * 扫描指定的监控目录是否产生了新文件或者文件是否被追加了内容
   */
  public List updateTailFiles (boolean skipToEnd) throws IOException {
    updateTime = System.currentTimeMillis();
    List updatedInodes = Lists.newArrayList();

    for (TaildirMatcher taildir : taildirCache) {
      Map headers = headerTable.row(taildir.getFileGroup());

      //遍历所有匹配的文件
      for (File f : taildir.getMatchingFiles()) {
        //得到本地文件的inode(储存文件元信息的区域就叫做inode，inode包含除了文件名以外的所有文件信息)
        //文件由唯一的inode，不论文件是否重命名，inode不变
        long inode = getInode(f);
        //tailFiles是一个Map，以inode为key，以TailFile为value
        //第一次遍历，此inode对应的Map项肯定不存在
        TailFile tf = tailFiles.get(inode);
        //源码中导致文件重命名后被重新读取的罪魁祸首
        //当文件重命名后，!tf.getPath().equals(f.getAbsolutePath())为True，那么就会创建新的TailFile，然后覆盖Map中原有的key-value对
        //TODO 源码修改处
        // if (tf == null || !tf.getPath().equals(f.getAbsolutePath())) {
        if (tf == null) {
          //如果Map中对应文件为空，那么就创建一个TailFile对象
          //skipToEnd可配置，决定是否从文件开始位置读取数据还是直接跳到文件结尾
          long startPos = skipToEnd ? f.length() : 0;
          //openFile中根据传入的参数new了一个新的TailFile
          tf = openFile(f, headers, inode, startPos);
        } else {
          //不为空时进入
          //如果文件重命名则进入此分支，由于是对于源码的修改导致重命名后进入，必须再次修改源码以处理重命名情况
          //判断此文件的更新时间是否比Map中存储的文件更新时间要新
          boolean updated = tf.getLastUpdated() < f.lastModified();
          if (updated) {
          //如果Map含有对应项，但是得到的tf中封装的文件为null，需要重新创建tf
            if (tf.getRaf() == null) {
              tf = openFile(f, headers, inode, tf.getPos());
            }

          // 如果Map中记录的读取位置Pos已经超过了文件长度，那么设置Map中的Pos值为0，即重新从0开始
            if (f.length() < tf.getPos()) {
              logger.info("Pos " + tf.getPos() + " is larger than file size! "
                  + "Restarting from pos 0, file: " + tf.getPath() + ", inode: " + inode);
              tf.updatePos(tf.getPath(), inode, 0);
            }
          }

          //重命名后，Map中的文件名还是老的文件名，因此使用openFIle重新创建TailFile用来替换原数据
          //TODO 源码修改处
          if (!tf.getPath().equals(f.getAbsolutePath())) {
            tf = openFile(f, headers, inode, tf.getPos());
          }
          //modify by zhangpeng end

          tf.setNeedTail(updated);
        }
        //将inode及其对应的tf加入Map中
        tailFiles.put(inode, tf);
        updatedInodes.add(inode);
      }
    }
    return updatedInodes;
  }

9、ReliableTaildirEventReader.java 文件loadPositionFile方法 源码修改

修改处，标有TODO，有一处修改

/**
   * Load a position file which has the last read position of each file.
   * 加载并解析记录了每个文件最新读取位置的position file
   * If the position file exists, update tailFiles mapping.
   * 如果position file存在则更新tailFiles映射
   */
    public void loadPositionFile(String filePath) {
    Long inode, pos;
    String path;
    FileReader fr = null;
    JsonReader jr = null;
    //对position file进行读取和解析
    try {
      fr = new FileReader(filePath);
      jr = new JsonReader(fr);
      jr.beginArray();
      while (jr.hasNext()) {
        inode = null;
        pos = null;
        path = null;
        jr.beginObject();
        while (jr.hasNext()) {
          switch (jr.nextName()) {
            case "inode":
              inode = jr.nextLong();
              break;
            case "pos":
              pos = jr.nextLong();
              break;
            case "file":
              path = jr.nextString();
              break;
          }
        }
        jr.endObject();

        for (Object v : Arrays.asList(inode, pos, path)) {
          Preconditions.checkNotNull(v, "Detected missing value in position file. "
              + "inode: " + inode + ", pos: " + pos + ", path: " + path);
        }
        //判断position file中的inode是否存在于TailFile Map中
        TailFile tf = tailFiles.get(inode);
        //根据对updatePos的分析，当出现重命名时，position file中的path项对应的文件名是旧文件名，而通过updateTailFiles()已经将Map中的文件名更新成了重命名后的文件名
        //因此，为了updatePos能够顺利更新pos，应该传入tf.getPath()，即新文件名，tailfile与tailfile自身的文件名的比较必然是相等的
        //TODO 源码修改处
        //if (tf != null && tf.updatePos(path, inode, pos)) {
        if (tf != null && tf.updatePos(tf.getPath(), inode, pos)) {
          tailFiles.put(inode, tf);
        } else {
          logger.info("Missing file: " + path + ", inode: " + inode + ", pos: " + pos);
        }
      }
      jr.endArray();
    } catch (FileNotFoundException e) {
      logger.info("File not found: " + filePath + ", not updating position");
    } catch (IOException e) {
      logger.error("Failed loading positionFile: " + filePath, e);
    } finally {
      try {
        if (fr != null) fr.close();
        if (jr != null) jr.close();
      } catch (IOException e) {
        logger.error("Error: " + e.getMessage(), e);
      }
    }
  }

10、pom.xml

    4.0.0

    com.djt.flume
    taildirsource
    1.0-SNAPSHOT

    
        UTF-8
    

    
        
            org.apache.flume
            flume-ng-core
            1.6.0
            provided
        
    

    
        flume-taildirsource
        
            
                org.apache.maven.plugins
                maven-resources-plugin
                2.6
                
                    ${project.build.sourceEncoding}
                
            

            
                org.apache.maven.plugins
                maven-compiler-plugin
                3.0
                
                    1.7
                    1.7
                    ${project.build.sourceEncoding}
                
            

        
    

11、maven package 打包

12、移植

将flume1.7打包好的源码放到flume1.6/lib目录下即可使用

13、Flume 应用案例 数据采集

a1.sources = r1
a1.channels = c1
a1.sinkgroups = g1
a1.sinks = k1 k2

a1.sources.r1.type = com.victor.flume.source.TaildirSource
a1.sources.r1.channels = c1
#TaildirSource 监控 nio 和 文件元数据信息 放到position.json  通过position去维护map
a1.sources.r1.positionFile = /opt/modules/flume/checkpoint/behavior/taildir_position.json
a1.sources.r1.filegroups = f1
##监控目录
a1.sources.r1.filegroups.f1 = /opt/modules/apache-tomcat-7.0.72-1/logs/OnlineStatistic/victor.log
a1.sources.r1.fileHeader = true

a1.channels.c1.type = file
a1.channels.c1.checkpointDir = /opt/modules/flume/checkpoint/behavior
a1.channels.c1.dataDirs = /opt/modules/flume/data/behavior/
a1.channels.c1.maxFileSize = 104857600
a1.channels.c1.capacity = 90000000
a1.channels.c1.keep-alive = 60

a1.sinkgroups.g1.sinks = k1 k2
a1.sinkgroups.g1.processor.type = load_balance
a1.sinkgroups.g1.processor.backoff = true
a1.sinkgroups.g1.processor.selector = round_robin
a1.sinkgroups.g1.processor.selector.maxTimeOut=10000

a1.sinks.k1.type = avro
a1.sinks.k1.channel = c1
a1.sinks.k1.batchSize = 1
a1.sinks.k1.hostname = hadoop102
a1.sinks.k1.port = 1234

a1.sinks.k2.type = avro
a1.sinks.k2.channel = c1
a1.sinks.k2.batchSize = 1
a1.sinks.k2.hostname = hadoop103
a1.sinks.k2.port = 1234

14、Flume 应用案例 日志聚合

a1.sources = r1
a1.channels = c1
a1.sinks = k1

a1.sources.r1.type = avro
a1.sources.r1.channels = c1
a1.sources.r1.bind = 0.0.0.0
a1.sources.r1.port = 1234

a1.channels.c1.type = file
a1.channels.c1.checkpointDir = /opt/modules/flume/checkpoint/behavior_collect
a1.channels.c1.dataDirs = /opt/modules/flume/data/behavior_collect
a1.channels.c1.maxFileSize = 104857600
a1.channels.c1.capacity = 90000000
a1.channels.c1.keep-alive = 60

a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.topic = t-behavior
a1.sinks.k1.brokerList = hadoop102:9092,hadoop103:9092,hadoop104:9092
a1.sinks.k1.requiredAcks = 1
a1.sinks.k1.kafka.producer.type = sync
a1.sinks.k1.batchSize = 1
a1.sinks.k1.channel = c1