org.apache.cassandra.hadoop

Class AbstractColumnFamilyRecordWriter<K,Y>

java.lang.Object

org.apache.hadoop.mapreduce.RecordWriter<K,Y>

org.apache.cassandra.hadoop.AbstractColumnFamilyRecordWriter<K,Y>

All Implemented Interfaces:

org.apache.hadoop.mapred.RecordWriter<K,Y>

public abstract class AbstractColumnFamilyRecordWriter<K,Y>
extends org.apache.hadoop.mapreduce.RecordWriter<K,Y>
implements org.apache.hadoop.mapred.RecordWriter<K,Y>

The ColumnFamilyRecordWriter maps the output <key, value> pairs to a Cassandra column family. In particular, it applies all mutations in the value, which it associates with the key, and in turn the responsible endpoint.

Furthermore, this writer groups the mutations by the endpoint responsible for the rows being affected. This allows the mutations to be executed in parallel, directly to a responsible endpoint.

See Also:

ColumnFamilyOutputFormat

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	AbstractColumnFamilyRecordWriter.AbstractRangeClient<K> A client that runs in a threadpool and connects to the list of endpoints for a particular range.

Field Summary

Fields

Modifier and Type	Field and Description
protected long	batchThreshold
protected org.apache.hadoop.conf.Configuration	conf
protected org.apache.cassandra.thrift.ConsistencyLevel	consistencyLevel
protected Progressable	progressable
protected int	queueSize
protected RingCache	ringCache

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	AbstractColumnFamilyRecordWriter(org.apache.hadoop.conf.Configuration conf)

Method Summary

Methods

Modifier and Type	Method and Description
protected abstract void	close()
void	close(org.apache.hadoop.mapred.Reporter reporter) Deprecated.
void	close(org.apache.hadoop.mapreduce.TaskAttemptContext context) Close this RecordWriter to future operations, but not before flushing out the batched mutations.

Methods inherited from class org.apache.hadoop.mapreduce.RecordWriter

write

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.hadoop.mapred.RecordWriter

write

Field Detail

conf

protected final org.apache.hadoop.conf.Configuration conf

ringCache

protected final RingCache ringCache

queueSize

protected final int queueSize

batchThreshold

protected final long batchThreshold

consistencyLevel

protected final org.apache.cassandra.thrift.ConsistencyLevel consistencyLevel

progressable

protected Progressable progressable

Constructor Detail

AbstractColumnFamilyRecordWriter

protected AbstractColumnFamilyRecordWriter(org.apache.hadoop.conf.Configuration conf)

Method Detail

close

public void close(org.apache.hadoop.mapreduce.TaskAttemptContext context)
           throws java.io.IOException,
                  java.lang.InterruptedException

Close this RecordWriter to future operations, but not before flushing out the batched mutations.

Specified by:

close in class org.apache.hadoop.mapreduce.RecordWriter<K,Y>

Parameters:

context - the context of the task

Throws:

java.io.IOException

java.lang.InterruptedException

close

@Deprecated
public void close(org.apache.hadoop.mapred.Reporter reporter)
           throws java.io.IOException

Deprecated.

Fills the deprecated RecordWriter interface for streaming.

Specified by:

close in interface org.apache.hadoop.mapred.RecordWriter<K,Y>

Throws:

java.io.IOException

close

protected abstract void close()
                       throws java.io.IOException

Throws:

java.io.IOException