/*
 * 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.db.compaction;

import java.util.*;
import java.util.function.Predicate;

import org.apache.cassandra.config.Config;
import org.apache.cassandra.db.Memtable;
import org.apache.cassandra.db.rows.UnfilteredRowIterator;

import com.google.common.base.Predicates;
import com.google.common.collect.Iterables;
import com.google.common.util.concurrent.RateLimiter;

import org.apache.cassandra.db.partitions.Partition;
import org.apache.cassandra.io.sstable.format.SSTableReader;
import org.apache.cassandra.io.util.FileDataInput;
import org.apache.cassandra.io.util.FileUtils;
import org.apache.cassandra.schema.CompactionParams.TombstoneOption;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import org.apache.cassandra.db.*;
import org.apache.cassandra.utils.AlwaysPresentFilter;
import org.apache.cassandra.utils.OverlapIterator;
import org.apache.cassandra.utils.concurrent.Refs;

import static org.apache.cassandra.db.lifecycle.SSTableIntervalTree.buildIntervals;

Manage compaction options.
/** * Manage compaction options. */
public class CompactionController implements AutoCloseable { private static final Logger logger = LoggerFactory.getLogger(CompactionController.class); private static final String NEVER_PURGE_TOMBSTONES_PROPERTY = Config.PROPERTY_PREFIX + "never_purge_tombstones"; static final boolean NEVER_PURGE_TOMBSTONES = Boolean.getBoolean(NEVER_PURGE_TOMBSTONES_PROPERTY); public final ColumnFamilyStore cfs; private final boolean compactingRepaired; // note that overlapIterator and overlappingSSTables will be null if NEVER_PURGE_TOMBSTONES is set - this is a // good thing so that noone starts using them and thinks that if overlappingSSTables is empty, there // is no overlap. private Refs<SSTableReader> overlappingSSTables; private OverlapIterator<PartitionPosition, SSTableReader> overlapIterator; private final Iterable<SSTableReader> compacting; private final RateLimiter limiter; private final long minTimestamp; final TombstoneOption tombstoneOption; final Map<SSTableReader, FileDataInput> openDataFiles = new HashMap<>(); public final int gcBefore; protected CompactionController(ColumnFamilyStore cfs, int maxValue) { this(cfs, null, maxValue); } public CompactionController(ColumnFamilyStore cfs, Set<SSTableReader> compacting, int gcBefore) { this(cfs, compacting, gcBefore, null, cfs.getCompactionStrategyManager().getCompactionParams().tombstoneOption()); } public CompactionController(ColumnFamilyStore cfs, Set<SSTableReader> compacting, int gcBefore, RateLimiter limiter, TombstoneOption tombstoneOption) { assert cfs != null; this.cfs = cfs; this.gcBefore = gcBefore; this.compacting = compacting; this.limiter = limiter; compactingRepaired = compacting != null && compacting.stream().allMatch(SSTableReader::isRepaired); this.tombstoneOption = tombstoneOption; this.minTimestamp = compacting != null && !compacting.isEmpty() // check needed for test ? compacting.stream().mapToLong(SSTableReader::getMinTimestamp).min().getAsLong() : 0; refreshOverlaps(); if (NEVER_PURGE_TOMBSTONES) logger.warn("You are running with -Dcassandra.never_purge_tombstones=true, this is dangerous!"); } public void maybeRefreshOverlaps() { if (NEVER_PURGE_TOMBSTONES) { logger.debug("not refreshing overlaps - running with -D{}=true", NEVER_PURGE_TOMBSTONES_PROPERTY); return; } if (ignoreOverlaps()) { logger.debug("not refreshing overlaps - running with ignoreOverlaps activated"); return; } for (SSTableReader reader : overlappingSSTables) { if (reader.isMarkedCompacted()) { refreshOverlaps(); return; } } } private void refreshOverlaps() { if (NEVER_PURGE_TOMBSTONES) return; if (this.overlappingSSTables != null) close(); if (compacting == null || ignoreOverlaps()) overlappingSSTables = Refs.tryRef(Collections.<SSTableReader>emptyList()); else overlappingSSTables = cfs.getAndReferenceOverlappingLiveSSTables(compacting); this.overlapIterator = new OverlapIterator<>(buildIntervals(overlappingSSTables)); } public Set<SSTableReader> getFullyExpiredSSTables() { return getFullyExpiredSSTables(cfs, compacting, overlappingSSTables, gcBefore, ignoreOverlaps()); }
Finds expired sstables works something like this; 1. find "global" minTimestamp of overlapping sstables, compacting sstables and memtables containing any non-expired data 2. build a list of fully expired candidates 3. check if the candidates to be dropped actually can be dropped (maxTimestamp < global minTimestamp) - if not droppable, remove from candidates 4. return candidates.
Params:
  • cfStore –
  • compacting – we take the drop-candidates from this set, it is usually the sstables included in the compaction
  • overlapping – the sstables that overlap the ones in compacting.
  • gcBefore –
  • ignoreOverlaps – don't check if data shadows/overlaps any data in other sstables
Returns:
/** * Finds expired sstables * * works something like this; * 1. find "global" minTimestamp of overlapping sstables, compacting sstables and memtables containing any non-expired data * 2. build a list of fully expired candidates * 3. check if the candidates to be dropped actually can be dropped {@code (maxTimestamp < global minTimestamp)} * - if not droppable, remove from candidates * 4. return candidates. * * @param cfStore * @param compacting we take the drop-candidates from this set, it is usually the sstables included in the compaction * @param overlapping the sstables that overlap the ones in compacting. * @param gcBefore * @param ignoreOverlaps don't check if data shadows/overlaps any data in other sstables * @return */
public static Set<SSTableReader> getFullyExpiredSSTables(ColumnFamilyStore cfStore, Iterable<SSTableReader> compacting, Iterable<SSTableReader> overlapping, int gcBefore, boolean ignoreOverlaps) { logger.trace("Checking droppable sstables in {}", cfStore); if (NEVER_PURGE_TOMBSTONES || compacting == null) return Collections.emptySet(); if (cfStore.getCompactionStrategyManager().onlyPurgeRepairedTombstones() && !Iterables.all(compacting, SSTableReader::isRepaired)) return Collections.emptySet(); if (ignoreOverlaps) { Set<SSTableReader> fullyExpired = new HashSet<>(); for (SSTableReader candidate : compacting) { if (candidate.getSSTableMetadata().maxLocalDeletionTime < gcBefore) { fullyExpired.add(candidate); logger.trace("Dropping overlap ignored expired SSTable {} (maxLocalDeletionTime={}, gcBefore={})", candidate, candidate.getSSTableMetadata().maxLocalDeletionTime, gcBefore); } } return fullyExpired; } List<SSTableReader> candidates = new ArrayList<>(); long minTimestamp = Long.MAX_VALUE; for (SSTableReader sstable : overlapping) { // Overlapping might include fully expired sstables. What we care about here is // the min timestamp of the overlapping sstables that actually contain live data. if (sstable.getSSTableMetadata().maxLocalDeletionTime >= gcBefore) minTimestamp = Math.min(minTimestamp, sstable.getMinTimestamp()); } for (SSTableReader candidate : compacting) { if (candidate.getSSTableMetadata().maxLocalDeletionTime < gcBefore) candidates.add(candidate); else minTimestamp = Math.min(minTimestamp, candidate.getMinTimestamp()); } for (Memtable memtable : cfStore.getTracker().getView().getAllMemtables()) minTimestamp = Math.min(minTimestamp, memtable.getMinTimestamp()); // At this point, minTimestamp denotes the lowest timestamp of any relevant // SSTable or Memtable that contains a constructive value. candidates contains all the // candidates with no constructive values. The ones out of these that have // (getMaxTimestamp() < minTimestamp) serve no purpose anymore. Iterator<SSTableReader> iterator = candidates.iterator(); while (iterator.hasNext()) { SSTableReader candidate = iterator.next(); if (candidate.getMaxTimestamp() >= minTimestamp) { iterator.remove(); } else { logger.trace("Dropping expired SSTable {} (maxLocalDeletionTime={}, gcBefore={})", candidate, candidate.getSSTableMetadata().maxLocalDeletionTime, gcBefore); } } return new HashSet<>(candidates); } public static Set<SSTableReader> getFullyExpiredSSTables(ColumnFamilyStore cfStore, Iterable<SSTableReader> compacting, Iterable<SSTableReader> overlapping, int gcBefore) { return getFullyExpiredSSTables(cfStore, compacting, overlapping, gcBefore, false); } public String getKeyspace() { return cfs.keyspace.getName(); } public String getColumnFamily() { return cfs.name; }
Params:
  • key –
Returns:a predicate for whether tombstones marked for deletion at the given time for the given partition are purgeable; we calculate this by checking whether the deletion time is less than the min timestamp of all SSTables containing his partition and not participating in the compaction. This means there isn't any data in those sstables that might still need to be suppressed by a tombstone at this timestamp.
/** * @param key * @return a predicate for whether tombstones marked for deletion at the given time for the given partition are * purgeable; we calculate this by checking whether the deletion time is less than the min timestamp of all SSTables * containing his partition and not participating in the compaction. This means there isn't any data in those * sstables that might still need to be suppressed by a tombstone at this timestamp. */
public Predicate<Long> getPurgeEvaluator(DecoratedKey key) { if (NEVER_PURGE_TOMBSTONES || !compactingRepaired()) return time -> false; overlapIterator.update(key); Set<SSTableReader> filteredSSTables = overlapIterator.overlaps(); Iterable<Memtable> memtables = cfs.getTracker().getView().getAllMemtables(); long minTimestampSeen = Long.MAX_VALUE; boolean hasTimestamp = false; for (SSTableReader sstable: filteredSSTables) { // if we don't have bloom filter(bf_fp_chance=1.0 or filter file is missing), // we check index file instead. if (sstable.getBloomFilter() instanceof AlwaysPresentFilter && sstable.getPosition(key, SSTableReader.Operator.EQ, false) != null || sstable.getBloomFilter().isPresent(key)) { minTimestampSeen = Math.min(minTimestampSeen, sstable.getMinTimestamp()); hasTimestamp = true; } } for (Memtable memtable : memtables) { Partition partition = memtable.getPartition(key); if (partition != null) { minTimestampSeen = Math.min(minTimestampSeen, partition.stats().minTimestamp); hasTimestamp = true; } } if (!hasTimestamp) return time -> true; else { final long finalTimestamp = minTimestampSeen; return time -> time < finalTimestamp; } } public void close() { if (overlappingSSTables != null) overlappingSSTables.release(); FileUtils.closeQuietly(openDataFiles.values()); openDataFiles.clear(); } public boolean compactingRepaired() { return !cfs.getCompactionStrategyManager().onlyPurgeRepairedTombstones() || compactingRepaired; } boolean provideTombstoneSources() { return tombstoneOption != TombstoneOption.NONE; } // caller must close iterators public Iterable<UnfilteredRowIterator> shadowSources(DecoratedKey key, boolean tombstoneOnly) { if (!provideTombstoneSources() || !compactingRepaired() || NEVER_PURGE_TOMBSTONES) return null; overlapIterator.update(key); return Iterables.filter(Iterables.transform(overlapIterator.overlaps(), reader -> getShadowIterator(reader, key, tombstoneOnly)), Predicates.notNull()); } @SuppressWarnings("resource") // caller to close private UnfilteredRowIterator getShadowIterator(SSTableReader reader, DecoratedKey key, boolean tombstoneOnly) { if (reader.isMarkedSuspect() || reader.getMaxTimestamp() <= minTimestamp || tombstoneOnly && !reader.mayHaveTombstones()) return null; RowIndexEntry<?> position = reader.getPosition(key, SSTableReader.Operator.EQ); if (position == null) return null; FileDataInput dfile = openDataFiles.computeIfAbsent(reader, this::openDataFile); return reader.simpleIterator(dfile, key, position, tombstoneOnly); }
Is overlapped sstables ignored Control whether or not we are taking into account overlapping sstables when looking for fully expired sstables. In order to reduce the amount of work needed, we look for sstables that can be dropped instead of compacted. As a safeguard mechanism, for each time range of data in a sstable, we are checking globally to see if all data of this time range is fully expired before considering to drop the sstable. This strategy can retain for a long time a lot of sstables on disk (see CASSANDRA-13418) so this option control whether or not this check should be ignored.
Returns:false by default
/** * Is overlapped sstables ignored * * Control whether or not we are taking into account overlapping sstables when looking for fully expired sstables. * In order to reduce the amount of work needed, we look for sstables that can be dropped instead of compacted. * As a safeguard mechanism, for each time range of data in a sstable, we are checking globally to see if all data * of this time range is fully expired before considering to drop the sstable. * This strategy can retain for a long time a lot of sstables on disk (see CASSANDRA-13418) so this option * control whether or not this check should be ignored. * * @return false by default */
protected boolean ignoreOverlaps() { return false; } private FileDataInput openDataFile(SSTableReader reader) { return limiter != null ? reader.openDataReader(limiter) : reader.openDataReader(); } }