org.apache.hadoop.hdfs.util
Class AtomicFileOutputStream

java.lang.Object
  java.io.OutputStream
      java.io.FilterOutputStream
          org.apache.hadoop.hdfs.util.AtomicFileOutputStream

All Implemented Interfaces:

Closeable, Flushable

public class AtomicFileOutputStream
extends FilterOutputStream

A FileOutputStream that has the property that it will only show up at its destination once it has been entirely written and flushed to disk. While being written, it will use a .tmp suffix. When the output stream is closed, it is flushed, fsynced, and will be moved into place, overwriting any file that already exists at that location. NOTE: on Windows platforms, it will not atomically replace the target file - instead the target file is deleted before this one is moved into place.

Field Summary

Fields inherited from class java.io.FilterOutputStream

out

Constructor Summary

AtomicFileOutputStream(File f)

Method Summary

void	abort() Close the atomic file, but do not "commit" the temporary file on top of the destination.
void	close()

Methods inherited from class java.io.FilterOutputStream

flush, write, write, write

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AtomicFileOutputStream

public AtomicFileOutputStream(File f)
                       throws FileNotFoundException

Throws:

FileNotFoundException

Method Detail

close

public void close()
           throws IOException

Specified by:

close in interface Closeable

Overrides:

close in class FilterOutputStream

Throws:

IOException

abort

public void abort()

Close the atomic file, but do not "commit" the temporary file on top of the destination. This should be used if there is a failure in writing.