Hadoop 2.4.1 setup Summary

First, let's take a look at the problem here, so as not to forget it. If you have time, you can get a detailed tutorial to play. The online tutorial is either an old version or an old version, or you just didn't say it to the point.

1. Disable IPv6 Configuration on the server.

2.if winutils.exe is not found, the system variable cannot be set with hadoop_home. environment variables can be directly set in the program, saving system configuration and making it easier.

System. setproperty ("hadoop. Home. dir", "XXXXXXXXX ");

3. Unable to connect to the server's HDFS Port

Check whether the host name is not bound to 127.0.0.1 or 127.0.1.1 in/etc/hosts. This is the main reason in Ubuntu. The host name in hosts must be bound to an Internet IP address.
Remove localhost from slaves, add the host names of all servers, one row, and bind the host names and corresponding Internet IP addresses of other servers to/etc/hosts.

If there is only one machine, we recommend that you directly store the host name. Otherwise, you need to modify the Host Name When copying data to other slave servers. Otherwise, "localhost" will overlap the slave servers below.

In some tutorials, IP addresses are used in XML, but I personally think that the host name is good, and then bind it to hosts. If the IP address is used, the IP address of the server will change, and there will be more to be changed. When the host name is used, you only need to change the host S.

The configuration in slaves is the key to datanode. Only the listed machines can have the datanode service.

4. After the HDFS server is connected to Windows, after the security mode is disabled, the operation can be read but cannot be written. This is because the current user name is used for communication by default. Is the permission verification disabled? Or change the current user name? Troublesome. just one line of code!

System. setproperty ("hadoop_user_name", "XXXXXXXX ");