Eclipse运行wordcount程序时报Connection timed out: no further information

Eclipse运行wordcount程序时报错,信息如下:

2020-08-15 16:12:32,580 INFO  [main] mapreduce.Job (Job.java:monitorAndPrintJob(1367)) -  map 0% reduce 0%
2020-08-15 16:12:52,739 WARN  [LocalJobRunner Map Task Executor #0] hdfs.BlockReaderFactory (BlockReaderFactory.java:getRemoteBlockReaderFromTcp(716)) - I/O error constructing remote block reader.
java.net.ConnectException: Connection timed out: no further information

Eclipse可以正常访问到hdfs,检查配置都没有问题,想到了用hadoop自带的例子去测试，结果一直卡在map 0% reduce 0%，后来更改mapred-site.xml为如下配置:

 <property>
        <name>mapreduce.job.tracker</name>
        <value>hdfs://hadoop01:9001</value>
        <final>true</final>
<property>

问题解决！

原因：mapreduce.framework.name 配置用yarn来进行计算时必须启动nodemanager,如果不使用yarn,进行mapreduce.job.tracker配置，也可以用MRv2来执行job，这样就不需要启动