Problems arise when a file is read hdfs record java.net.ConnectException: Connection refused

 

The company's hadoop cluster is built before colleagues and I (a white) to read files on hdfs when the spark shell, execute the following command

>>> word=sc.textFile("hdfs://localhost:9000/user/hadoop/test.txt")
>>> word.first()

Error: java.net.ConnectException: Call From hadoop / 133.0.123.130 to localhost: 9000 failed on connection exception: java.net.ConnectException: Connection refuse. Appears to be hdfs local server connection is a problem, I see the files on the following hdfs, discovery can view normal, indicating that the connection with hdfs local server, the communication is not a problem! Thought to want to, I have to change another way to read files on the hdfs

>>> word=sc.textFile("/user/hadoop/test.txt")
>>> word.first()

Because the file is read on hdfs spark default, so this way is possible, and found normal operation, under which the problem is clear, and is "localhost: 9000" out of the question, I look at the hadoop / etc / port disposed core-site.xml

​

Display port settings properly, it should be localhost corresponding IP address wrong! Then view the hosts file, found

​

Found localhost IP address corresponding to the local server IP addresses do not match, finally found a reason, I will read the instructions hdfs change the file:

>>> word=sc.textFile("hdfs://hadoop:9000/user/hadoop/test.txt")
>>> word.first()

The results normal.