Hadoop的简单控制台log分析
转自：http://blog.masr.in/geek/hadoop_mapreduce_log.html
以一个wordcount为例，详细讲解控制台输出的log信息，并通过改变jobconf的参数观察map reduce行为的变化。
首先把代码贴上来
<span style="background-color: rgb(0, 0, 0);">i</span><span style="background-color: rgb(255, 255, 255);">mport java.io.IOException;
import java.util.StringTokenizer;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IntWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.GenericOptionsParser;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
public class WordCount extends Configured implements Tool {
public static class TokenizerMapper extends
Mapper<Object, Text, Text, IntWritable> {
private final static IntWritable one = new IntWritable(1);
private Text word = new Text();
public void map(Object key, Text value, Context context)
throws IOException, InterruptedException {
StringTokenizer itr = new StringTokenizer(value.toString());
while (itr.hasMoreTokens()) {
String token = itr.nextToken();
if (token.isEmpty()) {
continue;
}
word.set(token);
context.write(word, one);
}
}
}
public static class IntSumReducer extends
Reducer<Text, IntWritable, Text, IntWritable> {
private IntWritable result = new IntWritable();
public void reduce(Text key, Iterable<IntWritable> values,
Context context) throws IOException, InterruptedException {
int sum = 0;
for (IntWritable val : values) {
sum += val.get();
}
result.set(sum);
context.write(key, result);
}
}
public int run(String[] args) throws Exception {
Configuration conf = new Configuration();
String[] otherArgs = new GenericOptionsParser(conf, args)
.getRemainingArgs();
if (otherArgs.length != 2) {
System.err.println("Usage: wordcount <in> <out>");
System.exit(2);
}
Job job = new Job(conf, "word count");
job.setJarByClass(WordCount.class);
job.setMapperClass(TokenizerMapper.class);
job.setCombinerClass(IntSumReducer.class);
job.setReducerClass(IntSumReducer.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(IntWritable.class);
FileInputFormat.addInputPath(job, new Path(otherArgs[0]));
FileOutputFormat.setOutputPath(job, new Path(otherArgs[1]));
System.exit(job.waitForCompletion(true) ? 0 : 1);
return 0;
}
public static void main(String[] args) throws Exception {
ToolRunner.run(new WordCount(), args);
}<span style="color:#ffffff;">
}</span></span>
首先我们来看以上这段代码的控制台结果，对应的解释嵌在结果里面
Oct 6, 2013 2:17:01 PM org.apache.hadoop.util.NativeCodeLoader <clinit> WARNING: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable Oct 6, 2013 2:17:01 PM org.apache.hadoop.mapred.JobClient copyAndConfigureFiles WARNING: No job jar file set. User classes may not be found. See JobConf(Class) or JobConf#setJar(String). Oct 6, 2013 2:17:01 PM org.apache.hadoop.mapreduce.lib.input.FileInputFormat listStatus //19是hadoop计算map reduce的时候总共要处理的文件的数量， //也就是FileInputFormat.addInputPath指定的文件夹里面的文件数量（如果指定的是文件夹的话）。 INFO: Total input paths to process : 19 Oct 6, 2013 2:17:01 PM org.apache.hadoop.io.compress.snappy.LoadSnappy <clinit> WARNING: Snappy native library not loaded Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.JobClient monitorAndPrintJob //1463375898_0001是本次map reduce的job_id号 INFO: Running job: job_local1463375898_0001 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.LocalJobRunner$Job run INFO: Waiting for map tasks Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.LocalJobRunner$Job$MapTaskRunnable run //这里开始跑第一个map，这里的attempt_local1463375898_0001_m_000000_0的意思是 //在jobid 1463375898_0001中的第0个map的第0次尝试。 INFO: Starting task: attempt_local1463375898_0001_m_000000_0 Oct 6, 2013 2:17:02 PM org.apache.hadoop.util.ProcessTree isSetsidSupported INFO: setsid exited with exit code 0 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.Task initialize INFO: Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@48e61a35 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.MapTask runNewMapper //显示我们将要处理的文件，这里表示第一个map处理该文件的从第0个字节到第0+3686382个字节的数据。 //有可能一个文件会被拆分给多个map处理。 INFO: Processing split: file:/home/coolmore/workspace/eclipse/HadoopStudy/data/in/hadoop_1.2.1.xml:0+3686382 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.MapTask$MapOutputBuffer <init> //一些mapreduce的参数 //io.sort.mb 指的是每个map存储中间结果的内存的最大size。如果超过这个值，数据就会spill到磁盘。 INFO: io.sort.mb = 100 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.MapTask$MapOutputBuffer <init> //这里796917760=996147200*0.8。 0.8指的是参数io.sort.spill.percent的值。 //这个值的意思是如果io.sort.mb为100MB，那么当内存中的数据达到80MB时就会开始把数据spill到磁盘， //并且在spill之前可能会执行combiner将结果集减小。 //在此之后会一边spill数据一边填充数据，类似于一个不断的生产和消费的过程。 INFO: data buffer = 796917760/996147200 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.MapTask$MapOutputBuffer <init> //2621440=3276800*0.8 INFO: record buffer = 2621440/3276800 Oct 6, 2013 2:17:02 PM org.apache.hadoop.mapred.MapTask$MapOutputBuffer flush //这里第一个map的数据在内存中足够放下，所以结果直接flush出来， //当然也是flush到磁盘上。在flush之前要在内存中进行sort操作。 //这里不存在mapper端的spill&merge的过程，因为数据量太小，只是spill生成了一个文件。 INFO: Starting flush of map output Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.JobClient monitorAndPrintJob INFO: map 0% reduce 0% Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.MapTask$MapOutputBuffer sortAndSpill INFO: Finished spill 0 //第一个map结束。 Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.Task done INFO: Task:attempt_local1463375898_0001_m_000000_0 is done. And is in the process of commiting Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.LocalJobRunner$Job statusUpdate INFO: Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.Task sendDone INFO: Task 'attempt_local1463375898_0001_m_000000_0' done. Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.LocalJobRunner$Job$MapTaskRunnable run INFO: Finishing task: attempt_local1463375898_0001_m_000000_0 Oct 6, 2013 2:17:03 PM org.apache.hadoop.mapred.LocalJobRunner$Job$MapTaskRunnable run //这里开始跑第二个map，这里的attempt_local1463375898_0001_m_000001_0的意思是 //在jobid attempt_local1463375898_0001中的第1个map的第0次尝试。接下来的步骤都是大同小异的，所以略过。 INFO: Starting task: attempt_local1463375898_0001_m_000001_0 。。。。。。 。。。。。。 。。。。。。 //最后一个map执行成功了。 INFO: Finishing task: attempt_local1463375898_0001_m_000018_0 Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.LocalJobRunner$Job run //所有的map执行成功了 INFO: Map task executor complete. Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Task initialize INFO: Using ResourceCalculatorPlugin : org.apache.hadoop.util.LinuxResourceCalculatorPlugin@79884a40 Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.LocalJobRunner$Job statusUpdate INFO: Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Merger$MergeQueue merge //这里其实到了reduce的部分，之前的19个map会生成19个文件，这些文件分别会被送到reduce中处理，并且这些文件都是已经排好序了。 //这里有一个参数很重要也就是io.sort.factor，意思是reduce和mapper做归并排序的最多处理的文件，本例中是10。 //因为我们总共有19个文件，所以reducer要先把这19个文件归并成10个。 INFO: Merging 19 sorted segments Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Merger$MergeQueue merge INFO: Merging 10 intermediate segments out of a total of 19 Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Merger$MergeQueue merge //10个文件总大小为7.9MB左右。 INFO: Down to the last merge-pass, with 10 segments left of total size: 7896329 bytes Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.LocalJobRunner$Job statusUpdate INFO: Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Task done //jobid为1463375898_0001的第1个reducer的第1次尝试完成了。 INFO: Task:attempt_local1463375898_0001_r_000000_0 is done. And is in the process of commiting Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.LocalJobRunner$Job statusUpdate INFO: Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Task commit INFO: Task attempt_local1463375898_0001_r_000000_0 is allowed to commit now Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter commitTask //将结果输出 INFO: Saved output of task 'attempt_local1463375898_0001_r_000000_0' to /home/coolmore/workspace/eclipse/HadoopStudy/data/out Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.LocalJobRunner$Job statusUpdate INFO: reduce > reduce Oct 6, 2013 2:17:12 PM org.apache.hadoop.mapred.Task sendDone INFO: Task 'attempt_local1463375898_0001_r_000000_0' done. Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.JobClient monitorAndPrintJob //进度条满了 INFO: map 100% reduce 100% Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.JobClient monitorAndPrintJob INFO: Job complete: job_local1463375898_0001 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log //把所有的counter的数据打出来把所有的counter的数据打出来，counter分成几个组，有 //FileInput Format counter //FileOutput Format Counters //MapReduce task counters //Filesystem counters //Job counters 几大类。 INFO: Counters: 20 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: File Output Format Counters Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Bytes Written=629401 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: File Input Format Counters Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Bytes Read=52879822 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: FileSystemCounters Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: FILE_BYTES_READ=672008497 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: FILE_BYTES_WRITTEN=101705449 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Map-Reduce Framework Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Map output materialized bytes=7896423 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Map input records=1252054 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Reduce shuffle bytes=0 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Spilled Records=784653 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Map output bytes=63190860 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: CPU time spent (ms)=0 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Total committed heap usage (bytes)=37932826624 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Combine input records=4020645 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: SPLIT_RAW_BYTES=2654 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Reduce input records=322311 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Reduce input groups=24731 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Combine output records=322311 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Physical memory (bytes) snapshot=0 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Reduce output records=24731 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Virtual memory (bytes) snapshot=0 Oct 6, 2013 2:17:13 PM org.apache.hadoop.mapred.Counters log INFO: Map output records=4020645
为了看得方便，我们把所有的counter都列出来。
FileInputFormat task counters:
- Bytes Read=52879822(BYTES_READ)
读取的input的数据文件的总大小为53MB左右。
FileOputFormat task counters:
- Bytes Written=629401(BYTES_WRITTEN)
最终在HDFS生成的文件的大小为629KB左右。
FileSystemCounters:
- FILE_BYTES_READ=672008497
累计读取本地磁盘的文件数据大小，map和reduce端有排序和归并，这些都需要占用io。
- FILE_BYTES_WRITTEN=101705449
累计写入本地磁盘的文件数据大小。
Map-Reduce Framework:
- Map input records=1252054(MAP_INPUT_RECORDS)
Map的输入的record数量，每次调用map函数的时候就会记一下。
- Map output records=4020645(MAP_OUTPUT_RECORDS)
Map输出的records的数量，每次调用context.write方法就会记录一下。
- Map output bytes=63190860(MAP_OUTPUT_BYTES)
非压缩的map的输出总大小，每次调用context.write方法就会记录一下。
- Combine input records=4020645(COMBINE_INPUT_RECORDS)
Combiner的处理的record的数量，因为combiner是紧接着mapper的，自然Combine input records=Map output records。
- Combine output records=322311(COMBINE_OUTPUT_RECORDS)
我们看到Combine input records=4020645，而Combine output records之后是322311，数据量大量减少了，提高了性能。
- Map output materialized bytes=7896423(MAP_OUTPUT_MATERIALIZED_BYTES)
据官方文档是说map真正输出到磁盘的文件的大小，如果map输出的文件是压缩的，那么改值就是压缩之后的值。 这里的值比Map output bytes小是因为combiner减小了数据量。
- Reduce input records=322311(REDUCE_INPUT_RECORDS)
Reducer所接受到的所有的record的数量，包含那些key是相同的record，每次迭代values的时候就会记一下。 这里 Combine output records=Reduce input records。
- Reduce input groups=24731(REDUCE_INPUT_GROUPS)
Reducer所接受到的所有的key的distinct值的数量，相当于是总的调用了多少次的reduce方法。
- Reduce output records=24731(REDUCE_OUTPUT_RECORDS)
这里Reduce output records=Reduce input groups，因为一个reducer方法只输出一行记录。
- Spilled Records=784653(SPILLED_RECORDS)
在map和reduce过程中splill到磁盘的record的数量