flume拦截器使用

log4j.properties配置：


log4j.rootLogger=INFO
log4j.category.com.besttone=INFO,flume
log4j.appender.flume = org.apache.flume.clients.log4jappender.Log4jAppender
log4j.appender.flume.Hostname = localhost
log4j.appender.flume.Port = 44444

log4j.appender.flume.UnsafeMode = true

需要将/opt/cloudera/parcels/CDH-5.0.0-1.cdh5.0.0.p0.47/lib/flume-ng/tools/flume-ng-log4jappender-1.4.0-cdh5.0.0-jar-with-dependencies.jar添加到classpath下。

然后可以写一个简单的测试类来测试一下：

 package com.besttone.flume;

 import java.util.Date;

 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;

 public class WriteLog {
     protected static final Log logger = LogFactory.getLog(WriteLog.class);

     /**
      * @param args
      * @throws InterruptedException
      */
     public static void main(String[] args) throws InterruptedException {
         // TODO Auto-generated method stub
         while (true) {
         //每隔两秒log输出一下当前系统时间戳
             logger.info(new Date().getTime());
             Thread.sleep(2000);
         }
     }
 }

然后写一个run.sh脚本运行这个类：

 #!/bin/bash
 jarlist=`ls ./lib/*.jar`
 CLASSPATH='./bin/'
 for jar in ${jarlist}
 do
    CLASSPATH=${CLASSPATH}:${jar}
 done
 echo ${CLASSPATH}

 java -classpath "$CLASSPATH" com.besttone.flume.WriteLog &

执行run.sh，将sink设置为logger,去flume的日志文件里去看，可以看到log4j的日志输出已经传输到了flume中：

2014-07-16 14:23:54,193 INFO org.apache.flume.sink.LoggerSink: Event: { headers:{flume.client.log4j.log.level=20000, flume.client.log4j.message.encoding=UTF8, flume.client.log4j.logger.name=com.besttone.flume.WriteLog, flume.client.log4j.timestamp=1405491834189} body: 31 34 30 35 34 39 31 38 33 34 31 38 39          1405491834189 }

对于flume拦截器,我的理解是：在app(应用程序日志)和 source 之间的，对app日志进行拦截处理的。也即在日志进入到source之前，对日志进行一些包装、清新过滤等等动作。

官方上提供的已有的拦截器有：

Timestamp Interceptor

Host Interceptor

Static Interceptor

Regex Filtering Interceptor

Regex Extractor Interceptor

像很多java的开源项目如springmvc中的拦截器一样，flume的拦截器也是chain形式的，可以对一个source指定多个拦截器，按先后顺序依次处理。

Timestamp Interceptor :在event的header中添加一个key叫：timestamp,value为当前的时间戳。这个拦截器在sink为hdfs 时很有用，后面会举例说到

Host Interceptor：在event的header中添加一个key叫：host,value为当前机器的hostname或者ip。

Static Interceptor:可以在event的header中添加自定义的key和value。

Regex Filtering Interceptor:通过正则来清洗或包含匹配的events。

Regex Extractor Interceptor：通过正则表达式来在header中添加指定的key,value则为正则匹配的部分

下面举例说明这些拦截器的用法，首先我们调整一下第一篇文章中的那个WriteLog类：

 public class WriteLog {
     protected static final Log logger = LogFactory.getLog(WriteLog.class);

     /**
      * @param args
      * @throws InterruptedException
      */
     public static void main(String[] args) throws InterruptedException {
         // TODO Auto-generated method stub
         while (true) {
             logger.info(new Date().getTime());
             logger.info("{\"requestTime\":"
                     + System.currentTimeMillis()
                     + ",\"requestParams\":{\"timestamp\":1405499314238,\"phone\":\"02038824941\",\"cardName\":\"测试商家名称\",\"provinceCode\":\"440000\",\"cityCode\":\"440106\"},\"requestUrl\":\"/reporter-api/reporter/reporter12/init.do\"}");
             Thread.sleep(2000);

         }
     }
 }

又多输出了一行日志信息，现在每次循环都会输出两行日志信息，第一行是一个时间戳信息，第二行是一行JSON格式的字符串信息。

接下来我们用regex_filter和 timestamp这两个拦截器来实现这样一个功能：

1 过滤掉LOG4J输出的第一行那个时间戳日志信息，只收集JSON格式的日志信息

2 将收集的日志信息保存到HDFS上，每天的日志保存到以该天命名的目录下面，如2014-7-25号的日志，保存到/flume/events/14-07-25目录下面。

修改后的flume.conf如下：

 tier1.sources=source1
 tier1.channels=channel1
 tier1.sinks=sink1

 tier1.sources.source1.type=avro
 tier1.sources.source1.bind=0.0.0.0
 tier1.sources.source1.port=44444
 tier1.sources.source1.channels=channel1

 tier1.sources.source1.interceptors=i1 i2
 tier1.sources.source1.interceptors.i1.type=regex_filter
 tier1.sources.source1.interceptors.i1.regex=\\{.*\\}
 tier1.sources.source1.interceptors.i2.type=timestamp

 tier1.channels.channel1.type=memory
 tier1.channels.channel1.capacity=10000
 tier1.channels.channel1.transactionCapacity=1000
 tier1.channels.channel1.keep-alive=30

 tier1.sinks.sink1.type=hdfs
 tier1.sinks.sink1.channel=channel1
 tier1.sinks.sink1.hdfs.path=hdfs://master68:8020/flume/events/%y-%m-%d
 tier1.sinks.sink1.hdfs.fileType=DataStream
 tier1.sinks.sink1.hdfs.writeFormat=Text
 tier1.sinks.sink1.hdfs.rollInterval=0
 tier1.sinks.sink1.hdfs.rollSize=10240
 tier1.sinks.sink1.hdfs.rollCount=0
 tier1.sinks.sink1.hdfs.idleTimeout=60

我们对source1添加了两个拦截器i1和i2,i1为regex_filter，过滤的正则为\\{.*\\},注意正则的写法用到了转义字符，不然source1无法启动，会报错。

i2为timestamp，在header中添加了一个timestamp的key,然后我们修改了sink1.hdfs.path在后面加上了/%y-%m-%d这一串字符，这一串字符要求event的header中必须有timestamp这个key,这就是为什么我们需要添加一个timestamp拦截器的原因，如果不添加这个拦截器，无法使用这样的占位符，会报错。还有很多占位符，请参考官方文档。

然后运行WriteLog,去hdfs上查看对应目录下面的文件，会发现内容只有JSON字符串的日志，与我们的功能描述一致。

先回想一下，spooldir source可以将文件名作为header中的key:basename写入到event的header当中去。试想一下，如果有一个拦截器可以拦截这个event,然后抽取header中这个key的值，将其拆分成3段，每一段都放入到header中，这样就可以实现那个需求了。

遗憾的是，flume没有提供可以拦截header的拦截器。不过有一个抽取body内容的拦截器：RegexExtractorInterceptor，看起来也很强大，以下是一个官方文档的示例：

If the Flume event body contained 1:2:3.4foobar5 and the following configuration was used


a1.sources.r1.interceptors.i1.regex = (\\d):(\\d):(\\d)
a1.sources.r1.interceptors.i1.serializers = s1 s2 s3
a1.sources.r1.interceptors.i1.serializers.s1.name = one
a1.sources.r1.interceptors.i1.serializers.s2.name = two
a1.sources.r1.interceptors.i1.serializers.s3.name = three
The extracted event will contain the same body but the following headers will have been added one=>1, two=>2, three=>3

大概意思就是，通过这样的配置，event body中如果有1:2:3.4foobar5 这样的内容，这会通过正则的规则抽取具体部分的内容，然后设置到header当中去。

于是决定打这个拦截器的主义，觉得只要把代码稍微改改，从拦截body改为拦截header中的具体key，就OK了。翻开源码，哎呀，很工整，改起来没难度，以下是我新增的一个拦截器：RegexExtractorExtInterceptor：

 package com.besttone.flume;

 import java.util.List;
 import java.util.Map;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;

 import org.apache.commons.lang.StringUtils;
 import org.apache.flume.Context;
 import org.apache.flume.Event;
 import org.apache.flume.interceptor.Interceptor;
 import org.apache.flume.interceptor.RegexExtractorInterceptorPassThroughSerializer;
 import org.apache.flume.interceptor.RegexExtractorInterceptorSerializer;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

 import com.google.common.base.Charsets;
 import com.google.common.base.Preconditions;
 import com.google.common.base.Throwables;
 import com.google.common.collect.Lists;

 /**
  * Interceptor that extracts matches using a specified regular expression and
  * appends the matches to the event headers using the specified serializers
  * Note that all regular expression matching occurs through Java's built in
  * java.util.regex package. Properties:
  *
  * regex: The regex to use
  *
  * serializers: Specifies the group the serializer will be applied to, and the
  * name of the header that will be added. If no serializer is specified for a
  * group the default {@link RegexExtractorInterceptorPassThroughSerializer} will
  * be used
  *
  * Sample config:
  *
  * agent.sources.r1.channels = c1
  *
  * agent.sources.r1.type = SEQ
  *
  * agent.sources.r1.interceptors = i1
  *
  * agent.sources.r1.interceptors.i1.type = REGEX_EXTRACTOR
  *
  * agent.sources.r1.interceptors.i1.regex = (WARNING)|(ERROR)|(FATAL)
  *
  * agent.sources.r1.interceptors.i1.serializers = s1 s2
  * agent.sources.r1.interceptors.i1.serializers.s1.type =
  * com.blah.SomeSerializer agent.sources.r1.interceptors.i1.serializers.s1.name
  * = warning agent.sources.r1.interceptors.i1.serializers.s2.type =
  * org.apache.flume.interceptor.RegexExtractorInterceptorTimestampSerializer
  * agent.sources.r1.interceptors.i1.serializers.s2.name = error
  * agent.sources.r1.interceptors.i1.serializers.s2.dateFormat = yyyy-MM-dd
  *
  *
  *
  *
  * Example 1:
  *
  * EventBody: 1:2:3.4foobar5 Configuration:
  * agent.sources.r1.interceptors.i1.regex = (\\d):(\\d):(\\d)
  *
  * agent.sources.r1.interceptors.i1.serializers = s1 s2 s3
  * agent.sources.r1.interceptors.i1.serializers.s1.name = one
  * agent.sources.r1.interceptors.i1.serializers.s2.name = two
  * agent.sources.r1.interceptors.i1.serializers.s3.name = three
  *
  * results in an event with the the following
  *
  * body: 1:2:3.4foobar5 headers: one=>1, two=>2, three=3
  *
  * Example 2:
  *
  * EventBody: 1:2:3.4foobar5
  *
  * Configuration: agent.sources.r1.interceptors.i1.regex = (\\d):(\\d):(\\d)
  *
  * agent.sources.r1.interceptors.i1.serializers = s1 s2
  * agent.sources.r1.interceptors.i1.serializers.s1.name = one
  * agent.sources.r1.interceptors.i1.serializers.s2.name = two
  *
  *
  * results in an event with the the following
  *
  * body: 1:2:3.4foobar5 headers: one=>1, two=>2
  *
  */
 public class RegexExtractorExtInterceptor implements Interceptor {

     static final String REGEX = "regex";
     static final String SERIALIZERS = "serializers";

     // 增加代码开始

     static final String EXTRACTOR_HEADER = "extractorHeader";
     static final boolean DEFAULT_EXTRACTOR_HEADER = false;
     static final String EXTRACTOR_HEADER_KEY = "extractorHeaderKey";

     // 增加代码结束

     private static final Logger logger = LoggerFactory
             .getLogger(RegexExtractorExtInterceptor.class);

     private final Pattern regex;
     private final List serializers;

     // 增加代码开始

     private final boolean extractorHeader;
     private final String extractorHeaderKey;

     // 增加代码结束

     private RegexExtractorExtInterceptor(Pattern regex,
             List serializers, boolean extractorHeader,
             String extractorHeaderKey) {
         this.regex = regex;
         this.serializers = serializers;
         this.extractorHeader = extractorHeader;
         this.extractorHeaderKey = extractorHeaderKey;
     }

     @Override
     public void initialize() {
         // NO-OP...
     }

     @Override
     public void close() {
         // NO-OP...
     }

     @Override
     public Event intercept(Event event) {
         String tmpStr;
         if(extractorHeader)
         {
             tmpStr = event.getHeaders().get(extractorHeaderKey);
         }
         else
         {
             tmpStr=new String(event.getBody(),
                     Charsets.UTF_8);
         }

         Matcher matcher = regex.matcher(tmpStr);
         Map headers = event.getHeaders();
         if (matcher.find()) {
             for (int group = 0, count = matcher.groupCount(); group < count; group++) {
                 int groupIndex = group + 1;
                 if (groupIndex > serializers.size()) {
                     if (logger.isDebugEnabled()) {
                         logger.debug(
                                 "Skipping group {} to {} due to missing serializer",
                                 group, count);
                     }
                     break;
                 }
                 NameAndSerializer serializer = serializers.get(group);
                 if (logger.isDebugEnabled()) {
                     logger.debug("Serializing {} using {}",
                             serializer.headerName, serializer.serializer);
                 }
                 headers.put(serializer.headerName, serializer.serializer
                         .serialize(matcher.group(groupIndex)));
             }
         }
         return event;
     }

     @Override
     public List intercept(List events) {
         List intercepted = Lists.newArrayListWithCapacity(events.size());
         for (Event event : events) {
             Event interceptedEvent = intercept(event);
             if (interceptedEvent != null) {
                 intercepted.add(interceptedEvent);
             }
         }
         return intercepted;
     }

     public static class Builder implements Interceptor.Builder {

         private Pattern regex;
         private List serializerList;

         // 增加代码开始

         private boolean extractorHeader;
         private String extractorHeaderKey;

         // 增加代码结束

         private final RegexExtractorInterceptorSerializer defaultSerializer = new RegexExtractorInterceptorPassThroughSerializer();

         @Override
         public void configure(Context context) {
             String regexString = context.getString(REGEX);
             Preconditions.checkArgument(!StringUtils.isEmpty(regexString),
                     "Must supply a valid regex string");

             regex = Pattern.compile(regexString);
             regex.pattern();
             regex.matcher("").groupCount();
             configureSerializers(context);

             // 增加代码开始
             extractorHeader = context.getBoolean(EXTRACTOR_HEADER,
                     DEFAULT_EXTRACTOR_HEADER);

             if (extractorHeader) {
                 extractorHeaderKey = context.getString(EXTRACTOR_HEADER_KEY);
                 Preconditions.checkArgument(
                         !StringUtils.isEmpty(extractorHeaderKey),
                         "必须指定要抽取内容的header key");
             }
             // 增加代码结束
         }

         private void configureSerializers(Context context) {
             String serializerListStr = context.getString(SERIALIZERS);
             Preconditions.checkArgument(
                     !StringUtils.isEmpty(serializerListStr),
                     "Must supply at least one name and serializer");

             String[] serializerNames = serializerListStr.split("\\s+");

             Context serializerContexts = new Context(
                     context.getSubProperties(SERIALIZERS + "."));

             serializerList = Lists
                     .newArrayListWithCapacity(serializerNames.length);
             for (String serializerName : serializerNames) {
                 Context serializerContext = new Context(
                         serializerContexts.getSubProperties(serializerName
                                 + "."));
                 String type = serializerContext.getString("type", "DEFAULT");
                 String name = serializerContext.getString("name");
                 Preconditions.checkArgument(!StringUtils.isEmpty(name),
                         "Supplied name cannot be empty.");

                 if ("DEFAULT".equals(type)) {
                     serializerList.add(new NameAndSerializer(name,
                             defaultSerializer));
                 } else {
                     serializerList.add(new NameAndSerializer(name,
                             getCustomSerializer(type, serializerContext)));
                 }
             }
         }

         private RegexExtractorInterceptorSerializer getCustomSerializer(
                 String clazzName, Context context) {
             try {
                 RegexExtractorInterceptorSerializer serializer = (RegexExtractorInterceptorSerializer) Class
                         .forName(clazzName).newInstance();
                 serializer.configure(context);
                 return serializer;
             } catch (Exception e) {
                 logger.error("Could not instantiate event serializer.", e);
                 Throwables.propagate(e);
             }
             return defaultSerializer;
         }

         @Override
         public Interceptor build() {
             Preconditions.checkArgument(regex != null,
                     "Regex pattern was misconfigured");
             Preconditions.checkArgument(serializerList.size() > 0,
                     "Must supply a valid group match id list");
             return new RegexExtractorExtInterceptor(regex, serializerList,
                     extractorHeader, extractorHeaderKey);
         }
     }

     static class NameAndSerializer {
         private final String headerName;
         private final RegexExtractorInterceptorSerializer serializer;

         public NameAndSerializer(String headerName,
                 RegexExtractorInterceptorSerializer serializer) {
             this.headerName = headerName;
             this.serializer = serializer;
         }
     }
 }

简单说明一下改动的内容：

增加了两个配置参数：

extractorHeader   是否抽取的是header部分，默认为false,即和原始的拦截器功能一致，抽取的是event body的内容

extractorHeaderKey 抽取的header的指定的key的内容，当extractorHeader为true时，必须指定该参数。

按照第八讲的方法，我们将该类打成jar包，作为flume的插件放到了/var/lib/flume-ng/plugins.d/RegexExtractorExtInterceptor/lib目录下，重新启动flume，将该拦截器加载到classpath中。

最终的flume.conf如下：

 tier1.sources=source1
 tier1.channels=channel1
 tier1.sinks=sink1
 tier1.sources.source1.type=spooldir
 tier1.sources.source1.spoolDir=/opt/logs
 tier1.sources.source1.fileHeader=true
 tier1.sources.source1.basenameHeader=true
 tier1.sources.source1.interceptors=i1
 tier1.sources.source1.interceptors.i1.type=com.besttone.flume.RegexExtractorExtInterceptor$Builder
 tier1.sources.source1.interceptors.i1.regex=(.*)\\.(.*)\\.(.*)
 tier1.sources.source1.interceptors.i1.extractorHeader=true
 tier1.sources.source1.interceptors.i1.extractorHeaderKey=basename
 tier1.sources.source1.interceptors.i1.serializers=s1 s2 s3
 tier1.sources.source1.interceptors.i1.serializers.s1.name=one
 tier1.sources.source1.interceptors.i1.serializers.s2.name=two
 tier1.sources.source1.interceptors.i1.serializers.s3.name=three
 tier1.sources.source1.channels=channel1
 tier1.sinks.sink1.type=hdfs
 tier1.sinks.sink1.channel=channel1
 tier1.sinks.sink1.hdfs.path=hdfs://master68:8020/flume/events/%{one}/%{three}
 tier1.sinks.sink1.hdfs.round=true
 tier1.sinks.sink1.hdfs.roundValue=10
 tier1.sinks.sink1.hdfs.roundUnit=minute
 tier1.sinks.sink1.hdfs.fileType=DataStream
 tier1.sinks.sink1.hdfs.writeFormat=Text
 tier1.sinks.sink1.hdfs.rollInterval=0
 tier1.sinks.sink1.hdfs.rollSize=10240
 tier1.sinks.sink1.hdfs.rollCount=0
 tier1.sinks.sink1.hdfs.idleTimeout=60
 tier1.channels.channel1.type=memory
 tier1.channels.channel1.capacity=10000
 tier1.channels.channel1.transactionCapacity=1000
 tier1.channels.channel1.keep-alive=30

我把source type改回了内置的spooldir，而不是上一讲自定义的source,然后添加了一个拦截器i1,type是自定义的拦截器：com.besttone.flume.RegexExtractorExtInterceptor$Builder,正则表达式按“.”分隔抽取三部分，分别放到header中的key:one,two,three当中去，即a.log.2014-07-31,通过拦截器后，在header当中就会增加三个key: one=a,two=log,three=2014-07-31。这时候我们在tier1.sinks.sink1.hdfs.path=hdfs://master68:8020/flume/events/%{one}/%{three}。

就实现了和前面第八讲一模一样的需求。

也可以看到，自定义拦截器的改动成本非常小，比自定义source小多了，我们这就增加了一个类，就实现了该功能。