Moving files in Hadoop using the Java API?

Question

I want to move files around in HDFS using the Java APIs. I cannot figure out a way to do this. The FileSystem class only seems to want to allow moving to and from the local file system.. but I want to keep them in HDFS and move them there.

Am I missing something basic? The only way I can figure to do it is to read it from the input stream and write it back out... and then delete the old copy (yuck).

thanks

Answer 1

Use FileSystem.rename():

public abstract boolean rename(Path src, Path dst) throws IOException

Renames Path src to Path dst. Can take place on local fs or remote DFS.

Parameters:
src - path to be renamed
dst - new path after rename
Returns:
true if rename is successful
Throws:
IOException - on failure

Answer 2

The java.nio.* approach may not work on HDFS always. So found the following solution that works.

Move files from one directory to another using org.apache.hadoop.fs.FileUtil.copy API

val fs = FileSystem.get(new Configuration())
        val conf = new org.apache.hadoop.conf.Configuration()
        val srcFs = FileSystem.get(new org.apache.hadoop.conf.Configuration())
        val dstFs = FileSystem.get(new org.apache.hadoop.conf.Configuration())
        val dstPath = new org.apache.hadoop.fs.Path(DEST_FILE_DIR)

        for (file <- fileList) {
          // The 5th parameter indicates whether source should be deleted or not
          FileUtil.copy(srcFs, file, dstFs, dstPath, true, conf)

Answer 3

I think the FileUtilts replaceFile would also solve the purpose. http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/fs/FileUtil.html#replaceFile(java.io.File, java.io.File)