JPS no Namenode after Hadoop started
Generally due to two or more than two times the format of the Namenode, there are two ways to solve:
1. Delete all data from Datanode
2. Modify the Namespaceid of each datanode (located in the/home/hdfs/data/current/version file) or modify the Namespaceid of Namenode (located in/home/hdfs/name/ Current/version file),
The aim is to agree on both.
But when viewed, the IDs of the two are the same,
Then look at the Namenode log file under/usr/local/hadoop/logs and find that the error is java.io.FileNotFoundException:/home/hadoop/hdfs/name/current/ VERSION (Permission denied)
After searching the internet, I found that it was/home/hadoop/hdfs/name/current/version's permission problem, so
[Email protected]:/usr/local/hadoop/bin$ sudo chmod-r 777/home/hadoop/hdfs
Then format: [Email protected]:/usr/local/hadoop/bin$ Hadoop Namenode-format
Start: [Email protected]:/usr/local/hadoop/bin$ start-all.sh
Jps:[email protected]:/usr/local/hadoop/bin$ JPS
6692 Jobtracker
6259 NameNode
6601 Secondarynamenode
6810 Jps
Seeing that Datanode and Tasktracker have not yet started, the following commands are used to start, but the discovery has actually started.
Jobtracker running as Process 6692. Stop it first.
192.168.1.3:tasktracker running as Process 4959. Stop it first.
192.168.1.4:tasktracker running as Process 5042. Stop it first.
[Email protected]:/usr/local/hadoop/bin$ JPS
6692 Jobtracker
6259 NameNode
6601 Secondarynamenode
7391 Jps
[Email protected]:/usr/local/hadoop/bin$ start-dfs.sh
Warning: $HADOOP _home is deprecated.
Namenode running as Process 6259. Stop it first.
192.168.1.3:datanode running as Process 4757. Stop it first.
192.168.1.4:datanode running as Process 4828. Stop it first.
192.168.1.2:secondarynamenode running as Process 6601. Stop it first.
Http://www.cnblogs.com/linjiqin/archive/2013/03/07/2948078.html
From Baidu know to see such a word,
The prompt says that this script is not recommended, use start-dfs.sh and start-mapred.sh to replace it. This means that the script author or maintainer also feels that the script may be problematic ... If you are interested, you can change it yourself ...
JPS no Namenode (RPM) after Hadoop startup