Package org.apache.cassandra.io.sstable.metadata

Class MetadataCollector

java.lang.Object

org.apache.cassandra.io.sstable.metadata.MetadataCollector

All Implemented Interfaces:

PartitionStatisticsCollector

public class MetadataCollector
extends java.lang.Object
implements PartitionStatisticsCollector

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	MetadataCollector.MinMaxIntTracker
static class	MetadataCollector.MinMaxLongTracker

Field Summary

Fields

Modifier and Type	Field	Description
protected com.clearspring.analytics.stream.cardinality.ICardinality	cardinality	Default cardinality estimation method is to use HyperLogLog++.
protected IntervalSet<CommitLogPosition>	commitLogIntervals
protected double	compressionRatio
protected EstimatedHistogram	estimatedCellPerPartitionCount
protected EstimatedHistogram	estimatedPartitionSize
protected StreamingTombstoneHistogramBuilder	estimatedTombstoneDropTime
protected boolean	hasLegacyCounterShards
protected MetadataCollector.MinMaxLongTracker	localDeletionTimeTracker
static double	NO_COMPRESSION_RATIO
protected int	sstableLevel
protected MetadataCollector.MinMaxLongTracker	timestampTracker
protected double	tokenSpaceCoverage
protected long	totalColumnsSet
protected long	totalRows
int	totalTombstones
protected MetadataCollector.MinMaxIntTracker	ttlTracker

Constructor Summary

Constructors

Constructor	Description
MetadataCollector(java.lang.Iterable<SSTableReader> sstables, ClusteringComparator comparator)
MetadataCollector(ClusteringComparator comparator)
MetadataCollector(ClusteringComparator comparator, java.util.UUID originatingHostId)

Method Summary

Modifier and Type	Method	Description
MetadataCollector	addCellPerPartitionCount()
MetadataCollector	addCellPerPartitionCount(long cellCount)
MetadataCollector	addCompressionRatio(long compressed, long uncompressed)	Ratio is compressed/uncompressed and it is if you have 1.x then compression isn't helping
MetadataCollector	addKey(java.nio.ByteBuffer key)
MetadataCollector	addPartitionSizeInBytes(long partitionSize)
MetadataCollector	commitLogIntervals(IntervalSet<CommitLogPosition> commitLogIntervals)
static StatsMetadata	defaultStatsMetadata()
java.util.Map<MetadataType,MetadataComponent>	finalizeMetadata(java.lang.String partitioner, double bloomFilterFPChance, long repairedAt, TimeUUID pendingRepair, boolean isTransient, SerializationHeader header, java.nio.ByteBuffer firstKey, java.nio.ByteBuffer lastKey)
void	release()	Release large memory objects while keeping metrics intact
MetadataCollector	sstableLevel(int sstableLevel)
MetadataCollector	tokenSpaceCoverage(double coverage)
void	update(DeletionTime dt)
void	update(LivenessInfo newInfo)
void	update(Cell<?> cell)
void	updateClusteringValues(Clustering<?> clustering)
void	updateClusteringValuesByBoundOrBoundary(ClusteringBoundOrBoundary<?> clusteringBoundOrBoundary)
void	updateColumnSetPerRow(long columnSetInRow)
void	updateHasLegacyCounterShards(boolean hasLegacyCounterShards)
void	updatePartitionDeletion(DeletionTime dt)

Methods inherited from class java.lang.Object

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

Field Detail

NO_COMPRESSION_RATIO

public static final double NO_COMPRESSION_RATIO

See Also:

Constant Field Values

estimatedPartitionSize

protected EstimatedHistogram estimatedPartitionSize

estimatedCellPerPartitionCount

protected EstimatedHistogram estimatedCellPerPartitionCount

commitLogIntervals

protected IntervalSet<CommitLogPosition> commitLogIntervals

timestampTracker

protected final MetadataCollector.MinMaxLongTracker timestampTracker

localDeletionTimeTracker

protected final MetadataCollector.MinMaxLongTracker localDeletionTimeTracker

ttlTracker

protected final MetadataCollector.MinMaxIntTracker ttlTracker

compressionRatio

protected double compressionRatio

estimatedTombstoneDropTime

protected StreamingTombstoneHistogramBuilder estimatedTombstoneDropTime

sstableLevel

protected int sstableLevel

hasLegacyCounterShards

protected boolean hasLegacyCounterShards

totalColumnsSet

protected long totalColumnsSet

totalRows

protected long totalRows

totalTombstones

public int totalTombstones

tokenSpaceCoverage

protected double tokenSpaceCoverage

cardinality

protected com.clearspring.analytics.stream.cardinality.ICardinality cardinality

Default cardinality estimation method is to use HyperLogLog++. Parameter here(p=13, sp=25) should give reasonable estimation while lowering bytes required to hold information. See CASSANDRA-5906 for detail.

Constructor Detail

MetadataCollector

public MetadataCollector(ClusteringComparator comparator)

MetadataCollector

public MetadataCollector(ClusteringComparator comparator,
                         java.util.UUID originatingHostId)

MetadataCollector

public MetadataCollector(java.lang.Iterable<SSTableReader> sstables,
                         ClusteringComparator comparator)

Method Detail

defaultStatsMetadata

public static StatsMetadata defaultStatsMetadata()

addKey

public MetadataCollector addKey(java.nio.ByteBuffer key)

addPartitionSizeInBytes

public MetadataCollector addPartitionSizeInBytes(long partitionSize)

addCellPerPartitionCount

public MetadataCollector addCellPerPartitionCount(long cellCount)

addCellPerPartitionCount

public MetadataCollector addCellPerPartitionCount()

addCompressionRatio

public MetadataCollector addCompressionRatio(long compressed,
                                             long uncompressed)

Ratio is compressed/uncompressed and it is if you have 1.x then compression isn't helping

update

public void update(LivenessInfo newInfo)

Specified by:

update in interface PartitionStatisticsCollector

update

public void update(Cell<?> cell)

Specified by:

update in interface PartitionStatisticsCollector

updatePartitionDeletion

public void updatePartitionDeletion(DeletionTime dt)

Specified by:

updatePartitionDeletion in interface PartitionStatisticsCollector

update

public void update(DeletionTime dt)

Specified by:

update in interface PartitionStatisticsCollector

updateColumnSetPerRow

public void updateColumnSetPerRow(long columnSetInRow)

Specified by:

updateColumnSetPerRow in interface PartitionStatisticsCollector

commitLogIntervals

public MetadataCollector commitLogIntervals(IntervalSet<CommitLogPosition> commitLogIntervals)

sstableLevel

public MetadataCollector sstableLevel(int sstableLevel)

tokenSpaceCoverage

public MetadataCollector tokenSpaceCoverage(double coverage)

updateClusteringValues

public void updateClusteringValues(Clustering<?> clustering)

updateClusteringValuesByBoundOrBoundary

public void updateClusteringValuesByBoundOrBoundary(ClusteringBoundOrBoundary<?> clusteringBoundOrBoundary)

updateHasLegacyCounterShards

public void updateHasLegacyCounterShards(boolean hasLegacyCounterShards)

Specified by:

updateHasLegacyCounterShards in interface PartitionStatisticsCollector

finalizeMetadata

public java.util.Map<MetadataType,MetadataComponent> finalizeMetadata(java.lang.String partitioner,
                                                                            double bloomFilterFPChance,
                                                                            long repairedAt,
                                                                            TimeUUID pendingRepair,
                                                                            boolean isTransient,
                                                                            SerializationHeader header,
                                                                            java.nio.ByteBuffer firstKey,
                                                                            java.nio.ByteBuffer lastKey)

release

public void release()

Release large memory objects while keeping metrics intact