/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.cassandra.io.sstable;
import java.io.*;
import java.nio.ByteBuffer;
import java.util.*;
import org.apache.cassandra.db.marshal.AbstractType;
import org.apache.cassandra.utils.ByteBufferUtil;
import org.apache.cassandra.utils.Pair;
import org.apache.cassandra.utils.StreamingHistogram;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.apache.cassandra.db.commitlog.ReplayPosition;
import org.apache.cassandra.io.util.FileUtils;
import org.apache.cassandra.utils.EstimatedHistogram;
/**
* Metadata for a SSTable.
* Metadata includes:
* - estimated row size histogram
* - estimated column count histogram
* - replay position
* - max column timestamp
* - max local deletion time
* - bloom filter fp chance
* - compression ratio
* - partitioner
* - generations of sstables from which this sstable was compacted, if any
* - tombstone drop time histogram
*
* An SSTableMetadata should be instantiated via the Collector, openFromDescriptor()
* or createDefaultInstance()
*/
public class SSTableMetadata
{
public static final double NO_BLOOM_FLITER_FP_CHANCE = -1.0;
public static final double NO_COMPRESSION_RATIO = -1.0;
public static final SSTableMetadataSerializer serializer = new SSTableMetadataSerializer();
public final EstimatedHistogram estimatedRowSize;
public final EstimatedHistogram estimatedColumnCount;
public final ReplayPosition replayPosition;
public final long minTimestamp;
public final long maxTimestamp;
public final int maxLocalDeletionTime;
public final double bloomFilterFPChance;
public final double compressionRatio;
public final String partitioner;
public final StreamingHistogram estimatedTombstoneDropTime;
public final int sstableLevel;
public final List<ByteBuffer> maxColumnNames;
public final List<ByteBuffer> minColumnNames;
private SSTableMetadata()
{
this(defaultRowSizeHistogram(),
defaultColumnCountHistogram(),
ReplayPosition.NONE,
Long.MIN_VALUE,
Long.MAX_VALUE,
Integer.MAX_VALUE,
NO_BLOOM_FLITER_FP_CHANCE,
NO_COMPRESSION_RATIO,
null,
defaultTombstoneDropTimeHistogram(),
0,
Collections.<ByteBuffer>emptyList(),
Collections.<ByteBuffer>emptyList());
}
private SSTableMetadata(EstimatedHistogram rowSizes,
EstimatedHistogram columnCounts,
ReplayPosition replayPosition,
long minTimestamp,
long maxTimestamp,
int maxLocalDeletionTime,
double bloomFilterFPChance,
double compressionRatio,
String partitioner,
StreamingHistogram estimatedTombstoneDropTime,
int sstableLevel,
List<ByteBuffer> minColumnNames,
List<ByteBuffer> maxColumnNames)
{
this.estimatedRowSize = rowSizes;
this.estimatedColumnCount = columnCounts;
this.replayPosition = replayPosition;
this.minTimestamp = minTimestamp;
this.maxTimestamp = maxTimestamp;
this.maxLocalDeletionTime = maxLocalDeletionTime;
this.bloomFilterFPChance = bloomFilterFPChance;
this.compressionRatio = compressionRatio;
this.partitioner = partitioner;
this.estimatedTombstoneDropTime = estimatedTombstoneDropTime;
this.sstableLevel = sstableLevel;
this.minColumnNames = minColumnNames;
this.maxColumnNames = maxColumnNames;
}
public static Collector createCollector(AbstractType<?> columnNameComparator)
{
return new Collector(columnNameComparator);
}
public static Collector createCollector(Collection<SSTableReader> sstables, AbstractType<?> columnNameComparator, int level)
{
Collector collector = new Collector(columnNameComparator);
collector.replayPosition(ReplayPosition.getReplayPosition(sstables));
collector.sstableLevel(level);
// Get the max timestamp of the precompacted sstables
// and adds generation of live ancestors
for (SSTableReader sstable : sstables)
{
collector.addAncestor(sstable.descriptor.generation);
for (Integer i : sstable.getAncestors())
{
if (new File(sstable.descriptor.withGeneration(i).filenameFor(Component.DATA)).exists())
collector.addAncestor(i);
}
}
return collector;
}
/**
* Used when updating sstablemetadata files with an sstable level
* @param metadata
* @param sstableLevel
* @return
*/
@Deprecated
public static SSTableMetadata copyWithNewSSTableLevel(SSTableMetadata metadata, int sstableLevel)
{
return new SSTableMetadata(metadata.estimatedRowSize,
metadata.estimatedColumnCount,
metadata.replayPosition,
metadata.minTimestamp,
metadata.maxTimestamp,
metadata.maxLocalDeletionTime,
metadata.bloomFilterFPChance,
metadata.compressionRatio,
metadata.partitioner,
metadata.estimatedTombstoneDropTime,
sstableLevel,
metadata.minColumnNames,
metadata.maxColumnNames);
}
static EstimatedHistogram defaultColumnCountHistogram()
{
// EH of 114 can track a max value of 2395318855, i.e., > 2B columns
return new EstimatedHistogram(114);
}
static EstimatedHistogram defaultRowSizeHistogram()
{
// EH of 150 can track a max value of 1697806495183, i.e., > 1.5PB
return new EstimatedHistogram(150);
}
static StreamingHistogram defaultTombstoneDropTimeHistogram()
{
return new StreamingHistogram(SSTable.TOMBSTONE_HISTOGRAM_BIN_SIZE);
}
/**
* @param gcBefore
* @return estimated droppable tombstone ratio at given gcBefore time.
*/
public double getEstimatedDroppableTombstoneRatio(int gcBefore)
{
long estimatedColumnCount = this.estimatedColumnCount.mean() * this.estimatedColumnCount.count();
if (estimatedColumnCount > 0)
{
double droppable = getDroppableTombstonesBefore(gcBefore);
return droppable / estimatedColumnCount;
}
return 0.0f;
}
/**
* Get the amount of droppable tombstones
* @param gcBefore the gc time
* @return amount of droppable tombstones
*/
public double getDroppableTombstonesBefore(int gcBefore)
{
return estimatedTombstoneDropTime.sum(gcBefore);
}
public static class Collector
{
protected EstimatedHistogram estimatedRowSize = defaultRowSizeHistogram();
protected EstimatedHistogram estimatedColumnCount = defaultColumnCountHistogram();
protected ReplayPosition replayPosition = ReplayPosition.NONE;
protected long minTimestamp = Long.MAX_VALUE;
protected long maxTimestamp = Long.MIN_VALUE;
protected int maxLocalDeletionTime = Integer.MIN_VALUE;
protected double compressionRatio = NO_COMPRESSION_RATIO;
protected Set<Integer> ancestors = new HashSet<Integer>();
protected StreamingHistogram estimatedTombstoneDropTime = defaultTombstoneDropTimeHistogram();
protected int sstableLevel;
protected List<ByteBuffer> minColumnNames = Collections.emptyList();
protected List<ByteBuffer> maxColumnNames = Collections.emptyList();
private final AbstractType<?> columnNameComparator;
private Collector(AbstractType<?> columnNameComparator)
{
this.columnNameComparator = columnNameComparator;
}
public void addRowSize(long rowSize)
{
estimatedRowSize.add(rowSize);
}
public void addColumnCount(long columnCount)
{
estimatedColumnCount.add(columnCount);
}
public void mergeTombstoneHistogram(StreamingHistogram histogram)
{
estimatedTombstoneDropTime.merge(histogram);
}
/**
* Ratio is compressed/uncompressed and it is
* if you have 1.x then compression isn't helping
*/
public void addCompressionRatio(long compressed, long uncompressed)
{
compressionRatio = (double) compressed/uncompressed;
}
public void updateMinTimestamp(long potentialMin)
{
minTimestamp = Math.min(minTimestamp, potentialMin);
}
public void updateMaxTimestamp(long potentialMax)
{
maxTimestamp = Math.max(maxTimestamp, potentialMax);
}
public void updateMaxLocalDeletionTime(int maxLocalDeletionTime)
{
this.maxLocalDeletionTime = Math.max(this.maxLocalDeletionTime, maxLocalDeletionTime);
}
public SSTableMetadata finalizeMetadata(String partitioner, double bloomFilterFPChance)
{
return new SSTableMetadata(estimatedRowSize,
estimatedColumnCount,
replayPosition,
minTimestamp,
maxTimestamp,
maxLocalDeletionTime,
bloomFilterFPChance,
compressionRatio,
partitioner,
estimatedTombstoneDropTime,
sstableLevel,
minColumnNames,
maxColumnNames);
}
public Collector estimatedRowSize(EstimatedHistogram estimatedRowSize)
{
this.estimatedRowSize = estimatedRowSize;
return this;
}
public Collector estimatedColumnCount(EstimatedHistogram estimatedColumnCount)
{
this.estimatedColumnCount = estimatedColumnCount;
return this;
}
public Collector replayPosition(ReplayPosition replayPosition)
{
this.replayPosition = replayPosition;
return this;
}
public Collector addAncestor(int generation)
{
this.ancestors.add(generation);
return this;
}
void update(long size, ColumnStats stats)
{
updateMinTimestamp(stats.minTimestamp);
/*
* The max timestamp is not always collected here (more precisely, row.maxTimestamp() may return Long.MIN_VALUE),
* to avoid deserializing an EchoedRow.
* This is the reason why it is collected first when calling ColumnFamilyStore.createCompactionWriter
* However, for old sstables without timestamp, we still want to update the timestamp (and we know
* that in this case we will not use EchoedRow, since CompactionControler.needsDeserialize() will be true).
*/
updateMaxTimestamp(stats.maxTimestamp);
updateMaxLocalDeletionTime(stats.maxLocalDeletionTime);
addRowSize(size);
addColumnCount(stats.columnCount);
mergeTombstoneHistogram(stats.tombstoneHistogram);
updateMinColumnNames(stats.minColumnNames);
updateMaxColumnNames(stats.maxColumnNames);
}
public Collector sstableLevel(int sstableLevel)
{
this.sstableLevel = sstableLevel;
return this;
}
public Collector updateMinColumnNames(List<ByteBuffer> minColumnNames)
{
if (minColumnNames.size() > 0)
this.minColumnNames = ColumnNameHelper.mergeMin(this.minColumnNames, minColumnNames, columnNameComparator);
return this;
}
public Collector updateMaxColumnNames(List<ByteBuffer> maxColumnNames)
{
if (maxColumnNames.size() > 0)
this.maxColumnNames = ColumnNameHelper.mergeMax(this.maxColumnNames, maxColumnNames, columnNameComparator);
return this;
}
}
public static class SSTableMetadataSerializer
{
private static final Logger logger = LoggerFactory.getLogger(SSTableMetadataSerializer.class);
public void serialize(SSTableMetadata sstableStats, Set<Integer> ancestors, DataOutput out) throws IOException
{
assert sstableStats.partitioner != null;
EstimatedHistogram.serializer.serialize(sstableStats.estimatedRowSize, out);
EstimatedHistogram.serializer.serialize(sstableStats.estimatedColumnCount, out);
ReplayPosition.serializer.serialize(sstableStats.replayPosition, out);
out.writeLong(sstableStats.minTimestamp);
out.writeLong(sstableStats.maxTimestamp);
out.writeInt(sstableStats.maxLocalDeletionTime);
out.writeDouble(sstableStats.bloomFilterFPChance);
out.writeDouble(sstableStats.compressionRatio);
out.writeUTF(sstableStats.partitioner);
out.writeInt(ancestors.size());
for (Integer g : ancestors)
out.writeInt(g);
StreamingHistogram.serializer.serialize(sstableStats.estimatedTombstoneDropTime, out);
out.writeInt(sstableStats.sstableLevel);
serializeMinMaxColumnNames(sstableStats.minColumnNames, sstableStats.maxColumnNames, out);
}
private void serializeMinMaxColumnNames(List<ByteBuffer> minColNames, List<ByteBuffer> maxColNames, DataOutput out) throws IOException
{
out.writeInt(minColNames.size());
for (ByteBuffer columnName : minColNames)
ByteBufferUtil.writeWithShortLength(columnName, out);
out.writeInt(maxColNames.size());
for (ByteBuffer columnName : maxColNames)
ByteBufferUtil.writeWithShortLength(columnName, out);
}
/**
* Used to serialize to an old version - needed to be able to update sstable level without a full compaction.
*
* @deprecated will be removed when it is assumed that the minimum upgrade-from-version is the version that this
* patch made it into
*
* @param sstableStats
* @param legacyDesc
* @param out
* @throws IOException
*/
@Deprecated
public void legacySerialize(SSTableMetadata sstableStats, Set<Integer> ancestors, Descriptor legacyDesc, DataOutput out) throws IOException
{
EstimatedHistogram.serializer.serialize(sstableStats.estimatedRowSize, out);
EstimatedHistogram.serializer.serialize(sstableStats.estimatedColumnCount, out);
ReplayPosition.serializer.serialize(sstableStats.replayPosition, out);
out.writeLong(sstableStats.minTimestamp);
out.writeLong(sstableStats.maxTimestamp);
if (legacyDesc.version.tracksMaxLocalDeletionTime)
out.writeInt(sstableStats.maxLocalDeletionTime);
if (legacyDesc.version.hasBloomFilterFPChance)
out.writeDouble(sstableStats.bloomFilterFPChance);
out.writeDouble(sstableStats.compressionRatio);
out.writeUTF(sstableStats.partitioner);
out.writeInt(ancestors.size());
for (Integer g : ancestors)
out.writeInt(g);
StreamingHistogram.serializer.serialize(sstableStats.estimatedTombstoneDropTime, out);
out.writeInt(sstableStats.sstableLevel);
if (legacyDesc.version.tracksMaxMinColumnNames)
serializeMinMaxColumnNames(sstableStats.minColumnNames, sstableStats.maxColumnNames, out);
}
/**
* deserializes the metadata
*
* returns a pair containing the part of the metadata meant to be kept-in memory and the part
* that should not.
*
* @param descriptor the descriptor
* @return a pair containing data that needs to be in memory and data that is potentially big and is not needed
* in memory
* @throws IOException
*/
public Pair<SSTableMetadata, Set<Integer>> deserialize(Descriptor descriptor) throws IOException
{
return deserialize(descriptor, true);
}
public Pair<SSTableMetadata, Set<Integer>> deserialize(Descriptor descriptor, boolean loadSSTableLevel) throws IOException
{
logger.debug("Load metadata for {}", descriptor);
File statsFile = new File(descriptor.filenameFor(SSTable.COMPONENT_STATS));
if (!statsFile.exists())
{
logger.debug("No sstable stats for {}", descriptor);
return Pair.create(new SSTableMetadata(), Collections.<Integer>emptySet());
}
DataInputStream in = new DataInputStream(new BufferedInputStream(new FileInputStream(statsFile)));
try
{
return deserialize(in, descriptor, loadSSTableLevel);
}
finally
{
FileUtils.closeQuietly(in);
}
}
public Pair<SSTableMetadata, Set<Integer>> deserialize(DataInputStream in, Descriptor desc) throws IOException
{
return deserialize(in, desc, true);
}
public Pair<SSTableMetadata, Set<Integer>> deserialize(DataInputStream in, Descriptor desc, boolean loadSSTableLevel) throws IOException
{
EstimatedHistogram rowSizes = EstimatedHistogram.serializer.deserialize(in);
EstimatedHistogram columnCounts = EstimatedHistogram.serializer.deserialize(in);
ReplayPosition replayPosition = ReplayPosition.serializer.deserialize(in);
long minTimestamp = in.readLong();
long maxTimestamp = in.readLong();
int maxLocalDeletionTime = desc.version.tracksMaxLocalDeletionTime ? in.readInt() : Integer.MAX_VALUE;
double bloomFilterFPChance = desc.version.hasBloomFilterFPChance ? in.readDouble() : NO_BLOOM_FLITER_FP_CHANCE;
double compressionRatio = in.readDouble();
String partitioner = in.readUTF();
int nbAncestors = in.readInt();
Set<Integer> ancestors = new HashSet<Integer>(nbAncestors);
for (int i = 0; i < nbAncestors; i++)
ancestors.add(in.readInt());
StreamingHistogram tombstoneHistogram = StreamingHistogram.serializer.deserialize(in);
int sstableLevel = 0;
if (loadSSTableLevel && in.available() > 0)
sstableLevel = in.readInt();
List<ByteBuffer> minColumnNames;
List<ByteBuffer> maxColumnNames;
if (desc.version.tracksMaxMinColumnNames)
{
int colCount = in.readInt();
minColumnNames = new ArrayList<ByteBuffer>(colCount);
for (int i = 0; i < colCount; i++)
{
minColumnNames.add(ByteBufferUtil.readWithShortLength(in));
}
colCount = in.readInt();
maxColumnNames = new ArrayList<ByteBuffer>(colCount);
for (int i = 0; i < colCount; i++)
{
maxColumnNames.add(ByteBufferUtil.readWithShortLength(in));
}
}
else
{
minColumnNames = Collections.emptyList();
maxColumnNames = Collections.emptyList();
}
return Pair.create(new SSTableMetadata(rowSizes,
columnCounts,
replayPosition,
minTimestamp,
maxTimestamp,
maxLocalDeletionTime,
bloomFilterFPChance,
compressionRatio,
partitioner,
tombstoneHistogram,
sstableLevel,
minColumnNames,
maxColumnNames), ancestors);
}
}
}