org.apache.hadoop.hdfs.server.datanode

Class ReplicaInPipeline

java.lang.Object

org.apache.hadoop.hdfs.protocol.Block

org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

org.apache.hadoop.hdfs.server.datanode.ReplicaInPipeline

All Implemented Interfaces:

Comparable<org.apache.hadoop.hdfs.protocol.Block>, org.apache.hadoop.hdfs.server.datanode.Replica, ReplicaInPipelineInterface, org.apache.hadoop.io.Writable, org.apache.hadoop.util.LightWeightGSet.LinkedElement

Direct Known Subclasses:

ReplicaBeingWritten

public class ReplicaInPipeline
extends org.apache.hadoop.hdfs.server.datanode.ReplicaInfo
implements ReplicaInPipelineInterface

This class defines a replica in a pipeline, which includes a persistent replica being written to by a dfs client or a temporary replica being replicated by a source datanode or being copied for the balancing purpose. The base class implements a temporary replica

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

ReplicaInfo.ReplicaDirInfo

Field Summary

Fields inherited from class org.apache.hadoop.hdfs.protocol.Block

BLOCK_FILE_PREFIX, blockFilePattern, METADATA_EXTENSION, metaFilePattern, metaOrBlockFilePattern

Constructor Summary

Constructors

Constructor and Description
ReplicaInPipeline(long blockId, long genStamp, FsVolumeSpi vol, File dir, long bytesToReserve) Constructor for a zero length replica
ReplicaInPipeline(ReplicaInPipeline from) Copy constructor.

Method Summary

Methods

Modifier and Type	Method and Description
boolean	attemptToSetWriter(Thread prevWriter, Thread newWriter) Attempt to set the writer to a new value.
OutputStream	createRestartMetaStream() Create an output stream to write restart metadata in case of datanode shutting down for quick restart.
ReplicaOutputStreams	createStreams(boolean isCreate, org.apache.hadoop.util.DataChecksum requestedChecksum) Create output streams for writing to this replica, one for block file and one for CRC file
boolean	equals(Object o)
long	getBytesAcked() Get the number of bytes acked
long	getBytesOnDisk() Get the number of bytes that have written to disk
long	getBytesReserved() Number of bytes reserved for this replica on disk.
ChunkChecksum	getLastChecksumAndDataLen() gets the last chunk checksum and the length of the block corresponding to that checksum
long	getOriginalBytesReserved() Number of bytes originally reserved for this replica.
HdfsServerConstants.ReplicaState	getState() Get the replica state
long	getVisibleLength() Get the number of bytes that are visible to readers
int	hashCode()
void	interruptThread()
void	releaseAllBytesReserved() Release any disk space reserved for this replica.
void	setBytesAcked(long bytesAcked) Set the number bytes that have acked
void	setLastChecksumAndDataLen(long dataLength, byte[] lastChecksum) store the checksum for the last chunk along with the data length
void	stopWriter(long xceiverStopTimeout) Interrupt the writing thread and wait until it dies
String	toString()

Methods inherited from class org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

breakHardLinksIfNeeded, getBlockFile, getMetaFile, getNext, getStorageUuid, getVolume, isOnTransientStorage, parseBaseDir, setDir, setNext

Methods inherited from class org.apache.hadoop.hdfs.protocol.Block

appendStringTo, compareTo, filename2id, getBlockId, getBlockId, getBlockName, getGenerationStamp, getGenerationStamp, getNumBytes, isBlockFilename, isMetaFilename, matchingIdAndGenStamp, metaToBlockFile, readFields, readId, set, setBlockId, setGenerationStamp, setNumBytes, toString, write, writeId

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface org.apache.hadoop.hdfs.server.datanode.ReplicaInPipelineInterface

setNumBytes

Methods inherited from interface org.apache.hadoop.hdfs.server.datanode.Replica

getBlockId, getGenerationStamp, getNumBytes, getStorageUuid, isOnTransientStorage

Constructor Detail

ReplicaInPipeline

public ReplicaInPipeline(long blockId,
                 long genStamp,
                 FsVolumeSpi vol,
                 File dir,
                 long bytesToReserve)

Constructor for a zero length replica

Parameters:

blockId - block id

genStamp - replica generation stamp

vol - volume where replica is located

dir - directory path where block and meta files are located

bytesToReserve - disk space to reserve for this replica, based on the estimated maximum block length.

ReplicaInPipeline

public ReplicaInPipeline(ReplicaInPipeline from)

Copy constructor.

Parameters:

from - where to copy from

Method Detail

getVisibleLength

public long getVisibleLength()

Description copied from interface: org.apache.hadoop.hdfs.server.datanode.Replica

Get the number of bytes that are visible to readers

Specified by:

getVisibleLength in interface org.apache.hadoop.hdfs.server.datanode.Replica

Returns:

the number of bytes that are visible to readers

getState

public HdfsServerConstants.ReplicaState getState()

Description copied from interface: org.apache.hadoop.hdfs.server.datanode.Replica

Get the replica state

Specified by:

getState in interface org.apache.hadoop.hdfs.server.datanode.Replica

Returns:

the replica state

getBytesAcked

public long getBytesAcked()

Description copied from interface: ReplicaInPipelineInterface

Get the number of bytes acked

Specified by:

getBytesAcked in interface ReplicaInPipelineInterface

Returns:

the number of bytes acked

setBytesAcked

public void setBytesAcked(long bytesAcked)

Description copied from interface: ReplicaInPipelineInterface

Set the number bytes that have acked

Specified by:

setBytesAcked in interface ReplicaInPipelineInterface

Parameters:

bytesAcked - number bytes acked

getBytesOnDisk

public long getBytesOnDisk()

Description copied from interface: org.apache.hadoop.hdfs.server.datanode.Replica

Get the number of bytes that have written to disk

Specified by:

getBytesOnDisk in interface org.apache.hadoop.hdfs.server.datanode.Replica

Returns:

the number of bytes that have written to disk

getBytesReserved

public long getBytesReserved()

Description copied from class: org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

Number of bytes reserved for this replica on disk.

Overrides:

getBytesReserved in class org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

getOriginalBytesReserved

public long getOriginalBytesReserved()

Description copied from class: org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

Number of bytes originally reserved for this replica. The actual reservation is adjusted as data is written to disk.

Overrides:

getOriginalBytesReserved in class org.apache.hadoop.hdfs.server.datanode.ReplicaInfo

Returns:

the number of bytes originally reserved for this replica.

releaseAllBytesReserved

public void releaseAllBytesReserved()

Description copied from interface: ReplicaInPipelineInterface

Release any disk space reserved for this replica.

Specified by:

releaseAllBytesReserved in interface ReplicaInPipelineInterface

setLastChecksumAndDataLen

public void setLastChecksumAndDataLen(long dataLength,
                             byte[] lastChecksum)

Description copied from interface: ReplicaInPipelineInterface

store the checksum for the last chunk along with the data length

Specified by:

setLastChecksumAndDataLen in interface ReplicaInPipelineInterface

Parameters:

dataLength - number of bytes on disk

lastChecksum - - checksum bytes for the last chunk

getLastChecksumAndDataLen

public ChunkChecksum getLastChecksumAndDataLen()

Description copied from interface: ReplicaInPipelineInterface

gets the last chunk checksum and the length of the block corresponding to that checksum

Specified by:

getLastChecksumAndDataLen in interface ReplicaInPipelineInterface

interruptThread

public void interruptThread()

equals

public boolean equals(Object o)

Overrides:

equals in class org.apache.hadoop.hdfs.protocol.Block

attemptToSetWriter

public boolean attemptToSetWriter(Thread prevWriter,
                         Thread newWriter)

Attempt to set the writer to a new value.

stopWriter

public void stopWriter(long xceiverStopTimeout)
                throws IOException

Interrupt the writing thread and wait until it dies

Throws:

IOException - the waiting is interrupted

hashCode

public int hashCode()

Overrides:

hashCode in class org.apache.hadoop.hdfs.protocol.Block

createStreams

public ReplicaOutputStreams createStreams(boolean isCreate,
                                 org.apache.hadoop.util.DataChecksum requestedChecksum)
                                   throws IOException

Description copied from interface: ReplicaInPipelineInterface

Create output streams for writing to this replica, one for block file and one for CRC file

Specified by:

createStreams in interface ReplicaInPipelineInterface

Parameters:

isCreate - if it is for creation

requestedChecksum - the checksum the writer would prefer to use

Returns:

output streams for writing

Throws:

IOException - if any error occurs

createRestartMetaStream

public OutputStream createRestartMetaStream()
                                     throws IOException

Description copied from interface: ReplicaInPipelineInterface

Create an output stream to write restart metadata in case of datanode shutting down for quick restart.

Specified by:

createRestartMetaStream in interface ReplicaInPipelineInterface

Returns:

output stream for writing.

Throws:

IOException - if any error occurs

toString

public String toString()

Overrides:

toString in class org.apache.hadoop.hdfs.server.datanode.ReplicaInfo