org.apache.lucene.store

Class MMapDirectory

java.lang.Object

org.apache.lucene.store.Directory

org.apache.lucene.store.BaseDirectory

org.apache.lucene.store.FSDirectory

org.apache.lucene.store.MMapDirectory

All Implemented Interfaces:

Closeable, AutoCloseable

public class MMapDirectory
extends FSDirectory

File-based Directory implementation that uses mmap for reading, and FSDirectory.FSIndexOutput for writing.

NOTE: memory mapping uses up a portion of the virtual memory address space in your process equal to the size of the file being mapped. Before using this class, be sure your have plenty of virtual address space, e.g. by using a 64 bit JRE, or a 32 bit JRE with indexes that are guaranteed to fit within the address space. On 32 bit platforms also consult MMapDirectory(File, LockFactory, int) if you have problems with mmap failing because of fragmented address space. If you get an OutOfMemoryException, it is recommended to reduce the chunk size, until it works.

Due to this bug in Sun's JRE, MMapDirectory's IndexInput.close() is unable to close the underlying OS file handle. Only when GC finally collects the underlying objects, which could be quite some time later, will the file handle be closed.

This will consume additional transient disk usage: on Windows, attempts to delete or overwrite the files will result in an exception; on other platforms, which typically have a "delete on last close" semantics, while such operations will succeed, the bytes are still consuming space on disk. For many applications this limitation is not a problem (e.g. if you have plenty of disk space, and you don't rely on overwriting files on Windows) but it's still an important limitation to be aware of.

This class supplies the workaround mentioned in the bug report (see setUseUnmap(boolean)), which may fail on non-Sun JVMs. It forcefully unmaps the buffer on close by using an undocumented internal cleanup functionality. UNMAP_SUPPORTED is true, if the workaround can be enabled (with no guarantees).

NOTE: Accessing this class either directly or indirectly from a thread while it's interrupted can close the underlying channel immediately if at the same time the thread is blocked on IO. The channel will remain closed and subsequent access to MMapDirectory will throw a ClosedChannelException.

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.lucene.store.Directory

Directory.IndexInputSlicer

Field Summary

Fields

Modifier and Type	Field and Description
static int	DEFAULT_MAX_BUFF Default max chunk size.
static boolean	UNMAP_SUPPORTED true, if this platform supports unmapping mmapped files.

Fields inherited from class org.apache.lucene.store.FSDirectory

DEFAULT_READ_CHUNK_SIZE

Constructor Summary

Constructors

Constructor and Description
MMapDirectory(File path) Create a new MMapDirectory for the named location and NativeFSLockFactory.
MMapDirectory(File path, LockFactory lockFactory) Create a new MMapDirectory for the named location.
MMapDirectory(File path, LockFactory lockFactory, int maxChunkSize) Create a new MMapDirectory for the named location, specifying the maximum chunk size used for memory mapping.

Method Summary

Modifier and Type	Method and Description
Directory.IndexInputSlicer	createSlicer(String name, IOContext context) Creates an Directory.IndexInputSlicer for the given file name.
int	getMaxChunkSize() Returns the current mmap chunk size.
boolean	getUseUnmap() Returns true, if the unmap workaround is enabled.
IndexInput	openInput(String name, IOContext context) Creates an IndexInput for the file with the given name.
void	setUseUnmap(boolean useUnmapHack) This method enables the workaround for unmapping the buffers from address space after closing IndexInput, that is mentioned in the bug report.

Methods inherited from class org.apache.lucene.store.FSDirectory

close, createOutput, deleteFile, fileExists, fileLength, getDirectory, getLockID, getReadChunkSize, listAll, listAll, open, open, setLockFactory, setReadChunkSize, sync, toString

Methods inherited from class org.apache.lucene.store.BaseDirectory

clearLock, getLockFactory, makeLock

Methods inherited from class org.apache.lucene.store.Directory

copy

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

DEFAULT_MAX_BUFF

public static final int DEFAULT_MAX_BUFF

Default max chunk size.

See Also:

MMapDirectory(File, LockFactory, int)

UNMAP_SUPPORTED

public static final boolean UNMAP_SUPPORTED

true, if this platform supports unmapping mmapped files.

Constructor Detail

MMapDirectory

public MMapDirectory(File path,
                     LockFactory lockFactory)
              throws IOException

Create a new MMapDirectory for the named location.

Parameters:

path - the path of the directory

lockFactory - the lock factory to use, or null for the default (NativeFSLockFactory);

Throws:

IOException - if there is a low-level I/O error

MMapDirectory

public MMapDirectory(File path)
              throws IOException

Create a new MMapDirectory for the named location and NativeFSLockFactory.

Parameters:

path - the path of the directory

Throws:

IOException - if there is a low-level I/O error

MMapDirectory

public MMapDirectory(File path,
                     LockFactory lockFactory,
                     int maxChunkSize)
              throws IOException

Create a new MMapDirectory for the named location, specifying the maximum chunk size used for memory mapping.

Parameters:

path - the path of the directory

lockFactory - the lock factory to use, or null for the default (NativeFSLockFactory);

maxChunkSize - maximum chunk size (default is 1 GiBytes for 64 bit JVMs and 256 MiBytes for 32 bit JVMs) used for memory mapping.

Especially on 32 bit platform, the address space can be very fragmented, so large index files cannot be mapped. Using a lower chunk size makes the directory implementation a little bit slower (as the correct chunk may be resolved on lots of seeks) but the chance is higher that mmap does not fail. On 64 bit Java platforms, this parameter should always be 1 << 30, as the address space is big enough.

Please note: The chunk size is always rounded down to a power of 2.

Throws:

IOException - if there is a low-level I/O error

Method Detail

setUseUnmap

public void setUseUnmap(boolean useUnmapHack)

This method enables the workaround for unmapping the buffers from address space after closing IndexInput, that is mentioned in the bug report. This hack may fail on non-Sun JVMs. It forcefully unmaps the buffer on close by using an undocumented internal cleanup functionality.

NOTE: Enabling this is completely unsupported by Java and may lead to JVM crashes if IndexInput is closed while another thread is still accessing it (SIGSEGV).

Throws:

IllegalArgumentException - if UNMAP_SUPPORTED is false and the workaround cannot be enabled.

getUseUnmap

public boolean getUseUnmap()

Returns true, if the unmap workaround is enabled.

See Also:

setUseUnmap(boolean)

getMaxChunkSize

public final int getMaxChunkSize()

Returns the current mmap chunk size.

See Also:

MMapDirectory(File, LockFactory, int)

openInput

public IndexInput openInput(String name,
                            IOContext context)
                     throws IOException

Creates an IndexInput for the file with the given name.

Specified by:

openInput in class Directory

Throws:

IOException

createSlicer

public Directory.IndexInputSlicer createSlicer(String name,
                                               IOContext context)
                                        throws IOException

Description copied from class: Directory

Creates an Directory.IndexInputSlicer for the given file name. IndexInputSlicer allows other Directory implementations to efficiently open one or more sliced IndexInput instances from a single file handle. The underlying file handle is kept open until the Directory.IndexInputSlicer is closed.

Must throw FileNotFoundException if the file does not exist (not java.nio.file.NoSuchFileException of Java 7).

Overrides:

createSlicer in class Directory

Throws:

IOException - if an IOException occurs