org.apache.cassandra.index.SecondaryIndexManager Maven / Gradle / Ivy
Show all versions of cassandra-all Show documentation
/*
* 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.index;
import java.lang.reflect.Constructor;
import java.util.*;
import java.util.concurrent.*;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.function.Function;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import com.google.common.annotations.VisibleForTesting;
import com.google.common.base.Joiner;
import com.google.common.base.Strings;
import com.google.common.collect.ImmutableSet;
import com.google.common.collect.Iterables;
import com.google.common.collect.Lists;
import com.google.common.collect.Maps;
import com.google.common.collect.Sets;
import com.google.common.primitives.Longs;
import com.google.common.util.concurrent.FutureCallback;
import com.google.common.util.concurrent.Futures;
import com.google.common.util.concurrent.ListenableFuture;
import com.google.common.util.concurrent.ListeningExecutorService;
import com.google.common.util.concurrent.MoreExecutors;
import com.google.common.util.concurrent.SettableFuture;
import org.apache.commons.lang3.StringUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.apache.cassandra.concurrent.JMXEnabledThreadPoolExecutor;
import org.apache.cassandra.concurrent.NamedThreadFactory;
import org.apache.cassandra.concurrent.StageManager;
import org.apache.cassandra.config.ColumnDefinition;
import org.apache.cassandra.config.DatabaseDescriptor;
import org.apache.cassandra.cql3.statements.IndexTarget;
import org.apache.cassandra.db.*;
import org.apache.cassandra.db.compaction.CompactionManager;
import org.apache.cassandra.db.filter.RowFilter;
import org.apache.cassandra.db.lifecycle.SSTableSet;
import org.apache.cassandra.db.lifecycle.View;
import org.apache.cassandra.db.partitions.*;
import org.apache.cassandra.db.rows.*;
import org.apache.cassandra.exceptions.InvalidRequestException;
import org.apache.cassandra.index.internal.CassandraIndex;
import org.apache.cassandra.index.transactions.*;
import org.apache.cassandra.io.sstable.format.SSTableReader;
import org.apache.cassandra.notifications.INotification;
import org.apache.cassandra.notifications.INotificationConsumer;
import org.apache.cassandra.notifications.SSTableAddedNotification;
import org.apache.cassandra.schema.IndexMetadata;
import org.apache.cassandra.schema.Indexes;
import org.apache.cassandra.service.pager.SinglePartitionPager;
import org.apache.cassandra.tracing.Tracing;
import org.apache.cassandra.transport.ProtocolVersion;
import org.apache.cassandra.transport.Server;
import org.apache.cassandra.utils.FBUtilities;
import org.apache.cassandra.utils.JVMStabilityInspector;
import org.apache.cassandra.utils.concurrent.OpOrder;
import org.apache.cassandra.utils.concurrent.Refs;
import static org.apache.cassandra.utils.ExecutorUtils.awaitTermination;
import static org.apache.cassandra.utils.ExecutorUtils.shutdown;
/**
* Handles the core maintenance functionality associated with indexes: adding/removing them to or from
* a table, (re)building during bootstrap or other streaming operations, flushing, reloading metadata
* and so on.
*
* The Index interface defines a number of methods which return {@code Callable}. These are primarily the
* management tasks for an index implementation. Most of them are currently executed in a blocking
* fashion via submission to SIM's blockingExecutor. This provides the desired behaviour in pretty
* much all cases, as tasks like flushing an index needs to be executed synchronously to avoid potentially
* deadlocking on the FlushWriter or PostFlusher. Several of these {@code Callable} returning methods on Index could
* then be defined with as void and called directly from SIM (rather than being run via the executor service).
* Separating the task defintion from execution gives us greater flexibility though, so that in future, for example,
* if the flush process allows it we leave open the possibility of executing more of these tasks asynchronously.
*
* The primary exception to the above is the Callable returned from Index#addIndexedColumn. This may
* involve a significant effort, building a new index over any existing data. We perform this task asynchronously;
* as it is called as part of a schema update, which we do not want to block for a long period. Building non-custom
* indexes is performed on the CompactionManager.
*
* This class also provides instances of processors which listen to updates to the base table and forward to
* registered Indexes the info required to keep those indexes up to date.
* There are two variants of these processors, each with a factory method provided by SIM:
* IndexTransaction: deals with updates generated on the regular write path.
* CleanupTransaction: used when partitions are modified during compaction or cleanup operations.
* Further details on their usage and lifecycles can be found in the interface definitions below.
*
* The bestIndexFor method is used at query time to identify the most selective index of those able
* to satisfy any search predicates defined by a ReadCommand's RowFilter. It returns a thin IndexAccessor object
* which enables the ReadCommand to access the appropriate functions of the Index at various stages in its lifecycle.
* e.g. the getEstimatedResultRows is required when StorageProxy calculates the initial concurrency factor for
* distributing requests to replicas, whereas a Searcher instance is needed when the ReadCommand is executed locally on
* a target replica.
*
* Finally, this class provides a clear and safe lifecycle to manage index builds, either full rebuilds via
* {@link this#rebuildIndexesBlocking(Set)} or builds of new sstables
* added via {@link org.apache.cassandra.notifications.SSTableAddedNotification}s, guaranteeing
* the following:
*
* - The initialization task and any subsequent successful (re)build mark the index as built.
* - If any (re)build operation fails, the index is not marked as built, and only another full rebuild can mark the
* index as built.
* - Full rebuilds cannot be run concurrently with other full or sstable (re)builds.
* - SSTable builds can always be run concurrently with any other builds.
*
*/
public class SecondaryIndexManager implements IndexRegistry, INotificationConsumer
{
private static final Logger logger = LoggerFactory.getLogger(SecondaryIndexManager.class);
// default page size (in rows) when rebuilding the index for a whole partition
public static final int DEFAULT_PAGE_SIZE = 10000;
/**
* All registered indexes.
*/
private final Map indexes = Maps.newConcurrentMap();
/**
* The indexes that had a build failure.
*/
private final Set needsFullRebuild = Sets.newConcurrentHashSet();
/**
* The indexes that are available for querying.
*/
private final Set queryableIndexes = Sets.newConcurrentHashSet();
/**
* The count of pending index builds for each index.
*/
private final Map inProgressBuilds = Maps.newConcurrentMap();
// executes tasks returned by Indexer#addIndexColumn which may require index(es) to be (re)built
private static final ListeningExecutorService asyncExecutor = MoreExecutors.listeningDecorator(
new JMXEnabledThreadPoolExecutor(1,
StageManager.KEEPALIVE,
TimeUnit.SECONDS,
new LinkedBlockingQueue<>(),
new NamedThreadFactory("SecondaryIndexManagement"),
"internal"));
// executes all blocking tasks produced by Indexers e.g. getFlushTask, getMetadataReloadTask etc
private static final ListeningExecutorService blockingExecutor = MoreExecutors.newDirectExecutorService();
/**
* The underlying column family containing the source data for these indexes
*/
public final ColumnFamilyStore baseCfs;
private final Keyspace keyspace;
public SecondaryIndexManager(ColumnFamilyStore baseCfs)
{
this.baseCfs = baseCfs;
this.keyspace = baseCfs.keyspace;
baseCfs.getTracker().subscribe(this);
}
/**
* Drops and adds new indexes associated with the underlying CF
*/
public void reload()
{
// figure out what needs to be added and dropped.
Indexes tableIndexes = baseCfs.metadata.getIndexes();
indexes.keySet()
.stream()
.filter(indexName -> !tableIndexes.has(indexName))
.forEach(this::removeIndex);
// we call add for every index definition in the collection as
// some may not have been created here yet, only added to schema
for (IndexMetadata tableIndex : tableIndexes)
addIndex(tableIndex, false);
}
private Future reloadIndex(IndexMetadata indexDef)
{
Index index = indexes.get(indexDef.name);
Callable reloadTask = index.getMetadataReloadTask(indexDef);
return reloadTask == null
? Futures.immediateFuture(null)
: blockingExecutor.submit(reloadTask);
}
@SuppressWarnings("unchecked")
private synchronized Future createIndex(IndexMetadata indexDef, boolean isNewCF)
{
final Index index = createInstance(indexDef);
index.register(this);
if (index.delayInitializationTask())
{
logger.info("Delay initialization task for index {}.{}.{}", baseCfs.keyspace.getName(), baseCfs.metadata.cfName, indexDef.name);
return Futures.immediateFuture(null);
}
markIndexesBuilding(ImmutableSet.of(index), true, isNewCF);
Callable initialBuildTask = null;
// if the index didn't register itself, we can probably assume that no initialization needs to happen
if (indexes.containsKey(indexDef.name))
{
try
{
initialBuildTask = index.getInitializationTask();
}
catch (Throwable t)
{
logAndMarkIndexesFailed(Collections.singleton(index), t);
throw t;
}
}
// if there's no initialization, just mark as built and return:
if (initialBuildTask == null)
{
markIndexBuilt(index, true);
return Futures.immediateFuture(null);
}
// otherwise run the initialization task asynchronously with a callback to mark it built or failed
final SettableFuture initialization = SettableFuture.create();
Futures.addCallback(asyncExecutor.submit(initialBuildTask), new FutureCallback()
{
@Override
public void onFailure(Throwable t)
{
logAndMarkIndexesFailed(Collections.singleton(index), t);
initialization.setException(t);
}
@Override
public void onSuccess(Object o)
{
markIndexBuilt(index, true);
initialization.set(o);
}
}, MoreExecutors.directExecutor());
return initialization;
}
/**
* Adds and builds a index
*
* @param indexDef the IndexMetadata describing the index
* @param isNewCF true if the index is added as part of a new table/columnfamily (i.e. loading a CF at startup),
* false for all other cases (i.e. newly added index)
*/
public synchronized Future addIndex(IndexMetadata indexDef, boolean isNewCF)
{
if (indexes.containsKey(indexDef.name))
return reloadIndex(indexDef);
else
return createIndex(indexDef, isNewCF);
}
public synchronized Future addIndex(IndexMetadata indexDef)
{
return addIndex(indexDef, true);
}
/**
* Checks if the specified index is queryable.
*
* @param index the index
* @return true if the specified index is registered, false otherwise
*/
public boolean isIndexQueryable(Index index)
{
return queryableIndexes.contains(index.getIndexMetadata().name);
}
/**
* Checks if the specified index has any running build task.
*
* @param indexName the index name
* @return {@code true} if the index is building, {@code false} otherwise
*/
@VisibleForTesting
public synchronized boolean isIndexBuilding(String indexName)
{
AtomicInteger counter = inProgressBuilds.get(indexName);
return counter != null && counter.get() > 0;
}
public synchronized void removeIndex(String indexName)
{
Index index = unregisterIndex(indexName);
if (null != index)
{
markIndexRemoved(indexName);
executeBlocking(index.getInvalidateTask(), null);
}
}
public Set getDependentIndexes(ColumnDefinition column)
{
if (indexes.isEmpty())
return Collections.emptySet();
Set dependentIndexes = new HashSet<>();
for (Index index : indexes.values())
if (index.dependsOn(column))
dependentIndexes.add(index.getIndexMetadata());
return dependentIndexes;
}
/**
* Called when dropping a Table
*/
public void markAllIndexesRemoved()
{
getBuiltIndexNames().forEach(this::markIndexRemoved);
}
/**
* Does a blocking full rebuild of the specifed indexes from all the sstables in the base table.
* Note also that this method of (re)building indexes:
* a) takes a set of index *names* rather than Indexers
* b) marks existing indexes removed prior to rebuilding
* c) fails if such marking operation conflicts with any ongoing index builds, as full rebuilds cannot be run
* concurrently
*
* @param indexNames the list of indexes to be rebuilt
*/
public void rebuildIndexesBlocking(Set indexNames)
{
try (ColumnFamilyStore.RefViewFragment viewFragment = baseCfs.selectAndReference(View.selectFunction(SSTableSet.CANONICAL));
Refs allSSTables = viewFragment.refs)
{
Set toRebuild = indexes.values().stream()
.filter(index -> indexNames.contains(index.getIndexMetadata().name))
.filter(Index::shouldBuildBlocking)
.collect(Collectors.toSet());
if (toRebuild.isEmpty())
{
logger.info("No defined indexes with the supplied names: {}", Joiner.on(',').join(indexNames));
return;
}
buildIndexesBlocking(allSSTables, toRebuild, true);
}
}
/**
* Does a full, blocking rebuild of the indexes specified by columns from the sstables.
* Caller must acquire and release references to the sstables used here.
* Note also that only this method of (re)building indexes:
* a) takes a set of index *names* rather than Indexers
* b) marks exsiting indexes removed prior to rebuilding
*
* @param sstables the data to build from
* @param indexNames the list of indexes to be rebuilt
*/
public void rebuildIndexesBlocking(Collection sstables, Set indexNames)
{
rebuildIndexesBlocking(1, sstables, indexNames);
}
public void rebuildIndexesBlocking(int indexThreads, Collection sstables, Set indexNames)
{
Set toRebuild = indexes.values().stream()
.filter(index -> indexNames.contains(index.getIndexMetadata().name))
.filter(Index::shouldBuildBlocking)
.collect(Collectors.toSet());
if (toRebuild.isEmpty())
{
logger.info("No defined indexes with the supplied names: {}", Joiner.on(',').join(indexNames));
return;
}
toRebuild.forEach(indexer -> markIndexRemoved(indexer.getIndexMetadata().name));
buildIndexesBlocking(indexThreads, sstables, toRebuild, true);
}
public void buildAllIndexesBlocking(Collection sstables)
{
buildIndexesBlocking(sstables, indexes.values()
.stream()
.filter(Index::shouldBuildBlocking)
.collect(Collectors.toSet()), true);
}
// For convenience, may be called directly from Index impls
public void buildIndexBlocking(Index index)
{
if (index.shouldBuildBlocking())
{
try (ColumnFamilyStore.RefViewFragment viewFragment = baseCfs.selectAndReference(View.selectFunction(SSTableSet.CANONICAL));
Refs sstables = viewFragment.refs)
{
buildIndexesBlocking(sstables, Collections.singleton(index), true);
markIndexBuilt(index.getIndexMetadata().name);
}
}
}
// Blocking build, for convenience, may be called directly from Index impls
public SettableFuture buildIndex(Index index)
{
if (index.shouldBuildBlocking())
{
try (ColumnFamilyStore.RefViewFragment viewFragment = baseCfs.selectAndReference(View.selectFunction(SSTableSet.CANONICAL));
Refs sstables = viewFragment.refs)
{
return buildIndex(1, sstables, index, true);
}
}
return null;
}
// Async build, for convenience, may be called directly from Index impls
public SettableFuture buildIndexAsync(Index index)
{
markIndexesBuilding(ImmutableSet.of(index), true, false);
final SettableFuture initialization = SettableFuture.create();
Futures.addCallback(buildIndex(index), new FutureCallback()
{
@Override
public void onFailure(Throwable t)
{
logAndMarkIndexesFailed(Collections.singleton(index), t);
initialization.setException(t);
}
@Override
public void onSuccess(Object o)
{
markIndexBuilt(index, true);
initialization.set(o);
}
}, MoreExecutors.directExecutor());
return initialization;
}
/**
* Non-blocking Index build on the compaction manager, for used by 2i implementations.
* @param indexThreads
* @param sstables
* @param indexes
* @param isFullRebuild
*/
private SettableFuture buildIndex(int indexThreads, Collection sstables, Index index, boolean isFullRebuild)
{
Set indexSingleton = Collections.singleton(index);
Index.IndexBuildingSupport buildingSupport = index.getBuildTaskSupport();
SecondaryIndexBuilder builder = buildingSupport.getIndexBuildTask(indexThreads, baseCfs, indexSingleton, sstables);
final SettableFuture build = SettableFuture.create();
Futures.addCallback(CompactionManager.instance.submitIndexBuild(builder), new FutureCallback()
{
@Override
public void onFailure(Throwable t)
{
logger.warn("Index build of " + getIndexNames(indexSingleton) + " failed:", t);
build.setException(t);
}
@Override
public void onSuccess(Object o)
{
logger.info("Index build of {} completed", getIndexNames(indexSingleton));
build.set(o);
}
});
return build;
}
/**
* Checks if the specified {@link ColumnFamilyStore} is a secondary index.
*
* @param cfs the ColumnFamilyStore to check.
* @return true if the specified ColumnFamilyStore is a secondary index,
* false otherwise.
*/
public static boolean isIndexColumnFamilyStore(ColumnFamilyStore cfs)
{
return isIndexColumnFamily(cfs.name);
}
/**
* Checks if the specified {@link ColumnFamilyStore} is the one secondary index.
*
* @param cfName the name of the ColumnFamilyStore to check.
* @return true if the specified ColumnFamilyStore is a secondary index,
* false otherwise.
*/
public static boolean isIndexColumnFamily(String cfName)
{
return cfName.contains(Directories.SECONDARY_INDEX_NAME_SEPARATOR);
}
/**
* Returns the parent of the specified {@link ColumnFamilyStore}.
*
* @param cfs the ColumnFamilyStore
* @return the parent of the specified ColumnFamilyStore
*/
public static ColumnFamilyStore getParentCfs(ColumnFamilyStore cfs)
{
String parentCfs = getParentCfsName(cfs.name);
return cfs.keyspace.getColumnFamilyStore(parentCfs);
}
/**
* Returns the parent name of the specified {@link ColumnFamilyStore}.
*
* @param cfName the ColumnFamilyStore name
* @return the parent name of the specified ColumnFamilyStore
*/
public static String getParentCfsName(String cfName)
{
assert isIndexColumnFamily(cfName);
return StringUtils.substringBefore(cfName, Directories.SECONDARY_INDEX_NAME_SEPARATOR);
}
/**
* Returns the index name
*
* @param cfs the ColumnFamilyStore
* @return the index name
*/
public static String getIndexName(ColumnFamilyStore cfs)
{
return getIndexName(cfs.name);
}
/**
* Returns the index name
*
* @param cfName the ColumnFamilyStore name
* @return the index name
*/
public static String getIndexName(String cfName)
{
assert isIndexColumnFamily(cfName);
return StringUtils.substringAfter(cfName, Directories.SECONDARY_INDEX_NAME_SEPARATOR);
}
/**
* Performs a blocking (re)indexing of the specified SSTables for the specified indexes.
*
* @param sstables the SSTables to be (re)indexed
* @param indexes the indexes to be (re)built for the specifed SSTables
* @param isFullRebuild True if this method is invoked as a full index rebuild, false otherwise
*/
@SuppressWarnings({ "unchecked" })
private void buildIndexesBlocking(Collection sstables, Set indexes, boolean isFullRebuild)
{
buildIndexesBlocking(1, sstables, indexes, isFullRebuild);
}
private void buildIndexesBlocking(int indexThreads, Collection sstables, Set indexes, boolean isFullRebuild)
{
if (indexes.isEmpty())
return;
// Mark all indexes as building: this step must happen first, because if any index can't be marked, the whole
// process needs to abort
markIndexesBuilding(indexes, isFullRebuild, false);
// Build indexes in a try/catch, so that any index not marked as either built or failed will be marked as failed:
final Set builtIndexes = new HashSet<>();
final Set unbuiltIndexes = new HashSet<>();
// Any exception thrown during index building that could be suppressed by the finally block
Exception accumulatedFail = null;
try
{
logger.info("Submitting index build of {} for data in {}",
indexes.stream().map(i -> i.getIndexMetadata().name).collect(Collectors.joining(",")),
sstables.stream().map(SSTableReader::toString).collect(Collectors.joining(",")));
// Group all building tasks
Map> byType = new HashMap<>();
for (Index index : indexes)
{
Set stored = byType.computeIfAbsent(index.getBuildTaskSupport(), i -> new HashSet<>());
stored.add(index);
}
// Schedule all index building tasks with a callback to mark them as built or failed
List> futures = new ArrayList<>(byType.size());
byType.forEach((buildingSupport, groupedIndexes) ->
{
SecondaryIndexBuilder builder = buildingSupport.getIndexBuildTask(indexThreads, baseCfs, groupedIndexes, sstables);
final SettableFuture build = SettableFuture.create();
Futures.addCallback(CompactionManager.instance.submitIndexBuild(builder), new FutureCallback()
{
@Override
public void onFailure(Throwable t)
{
logAndMarkIndexesFailed(groupedIndexes, t);
unbuiltIndexes.addAll(groupedIndexes);
build.setException(t);
}
@Override
public void onSuccess(Object o)
{
groupedIndexes.forEach(i -> markIndexBuilt(i, isFullRebuild));
logger.info("Index build of {} completed", getIndexNames(groupedIndexes));
builtIndexes.addAll(groupedIndexes);
build.set(o);
}
});
futures.add(build);
});
// Finally wait for the index builds to finish and flush the indexes that built successfully
FBUtilities.waitOnFutures(futures);
}
catch (Exception e)
{
accumulatedFail = e;
throw e;
}
finally
{
try
{
// Fail any indexes that couldn't be marked
Set failedIndexes = Sets.difference(indexes, Sets.union(builtIndexes, unbuiltIndexes));
if (!failedIndexes.isEmpty())
{
logAndMarkIndexesFailed(failedIndexes, accumulatedFail);
}
// Flush all built indexes with an aynchronous callback to log the success or failure of the flush
flushIndexesBlocking(builtIndexes, new FutureCallback()
{
String indexNames = StringUtils.join(builtIndexes.stream()
.map(i -> i.getIndexMetadata().name)
.collect(Collectors.toList()), ',');
@Override
public void onFailure(Throwable ignored)
{
logger.info("Index flush of {} failed", indexNames);
}
@Override
public void onSuccess(Object ignored)
{
logger.info("Index flush of {} completed", indexNames);
}
});
}
catch (Exception e)
{
if (accumulatedFail != null)
{
accumulatedFail.addSuppressed(e);
}
else
{
throw e;
}
}
}
}
private String getIndexNames(Set indexes)
{
List indexNames = indexes.stream()
.map(i -> i.getIndexMetadata().name)
.collect(Collectors.toList());
return StringUtils.join(indexNames, ',');
}
/**
* Marks the specified indexes as (re)building if:
* 1) There's no in progress rebuild of any of the given indexes.
* 2) There's an in progress rebuild but the caller is not a full rebuild.
*
* Otherwise, this method invocation fails, as it is not possible to run full rebuilds while other concurrent rebuilds
* are in progress. Please note this is checked atomically against all given indexes; that is, no index will be marked
* if even a single one fails.
*
* Marking an index as "building" practically means:
* 1) The index is removed from the "failed" set if this is a full rebuild.
* 2) The index is removed from the system keyspace built indexes; this only happens if this method is not invoked
* for a new table initialization, as in such case there's no need to remove it (it is either already not present,
* or already present because already built).
*
* Thread safety is guaranteed by having all methods managing index builds synchronized: being synchronized on
* the SecondaryIndexManager instance, it means all invocations for all different indexes will go through the same
* lock, but this is fine as the work done while holding such lock is trivial.
*
* {@link #markIndexBuilt(Index, boolean)} or {@link #markIndexFailed(Index)} should be always called after the
* rebuilding has finished, so that the index build state can be correctly managed and the index rebuilt.
*
* @param indexes the index to be marked as building
* @param isFullRebuild {@code true} if this method is invoked as a full index rebuild, {@code false} otherwise
* @param isNewCF {@code true} if this method is invoked when initializing a new table/columnfamily (i.e. loading a CF at startup),
* {@code false} for all other cases (i.e. newly added index)
*/
private synchronized void markIndexesBuilding(Set indexes, boolean isFullRebuild, boolean isNewCF)
{
String keyspaceName = baseCfs.keyspace.getName();
// First step is to validate against concurrent rebuilds; it would be more optimized to do everything on a single
// step, but we're not really expecting a very high number of indexes, and this isn't on any hot path, so
// we're favouring readability over performance
indexes.forEach(index ->
{
String indexName = index.getIndexMetadata().name;
AtomicInteger counter = inProgressBuilds.computeIfAbsent(indexName, ignored -> new AtomicInteger(0));
if (counter.get() > 0 && isFullRebuild)
throw new IllegalStateException(String.format("Cannot rebuild index %s as another index build for the same index is currently in progress.", indexName));
});
// Second step is the actual marking:
indexes.forEach(index ->
{
String indexName = index.getIndexMetadata().name;
AtomicInteger counter = inProgressBuilds.computeIfAbsent(indexName, ignored -> new AtomicInteger(0));
if (isFullRebuild)
needsFullRebuild.remove(indexName);
if (counter.getAndIncrement() == 0 && DatabaseDescriptor.isDaemonInitialized() && !isNewCF)
SystemKeyspace.setIndexRemoved(keyspaceName, indexName);
});
}
private synchronized void markIndexBuilt(Index index)
{
markIndexBuilt(index, true);
}
public synchronized void markIndexBuilt(String indexName)
{
Index index = indexes.get(indexName);
if (index != null)
markIndexBuilt(index, true);
}
/**
* Marks the specified indexes as (re)building if:
* 1) There's no in progress rebuild of any of the given indexes.
* 2) There's an in progress rebuild but the caller is not a full rebuild.
*
* Otherwise, this method invocation fails, as it is not possible to run full rebuilds while other concurrent rebuilds
* are in progress. Please note this is checked atomically against all given indexes; that is, no index will be marked
* if even a single one fails.
*
* Marking an index as "building" practically means:
* 1) The index is removed from the "failed" set if this is a full rebuild.
* 2) The index is removed from the system keyspace built indexes; this only happens if this method is not invoked
* for a new table initialization, as in such case there's no need to remove it (it is either already not present,
* or already present because already built).
*
* Thread safety is guaranteed by having all methods managing index builds synchronized: being synchronized on
* the SecondaryIndexManager instance, it means all invocations for all different indexes will go through the same
* lock, but this is fine as the work done while holding such lock is trivial.
*
* {@link #markIndexBuilt(Index, boolean)} or {@link #markIndexFailed(Index)} should be always called after the
* rebuilding has finished, so that the index build state can be correctly managed and the index rebuilt.
*
* @param indexes the index to be marked as building
* @param isFullRebuild {@code true} if this method is invoked as a full index rebuild, {@code false} otherwise
* @param isNewCF {@code true} if this method is invoked when initializing a new table/columnfamily (i.e. loading a CF at startup),
* {@code false} for all other cases (i.e. newly added index)
*/
private synchronized void markIndexBuilt(Index index, boolean isFullRebuild)
{
String indexName = index.getIndexMetadata().name;
if (isFullRebuild)
queryableIndexes.add(indexName);
AtomicInteger counter = inProgressBuilds.get(indexName);
if (counter != null)
{
assert counter.get() > 0;
if (counter.decrementAndGet() == 0)
{
inProgressBuilds.remove(indexName);
if (!needsFullRebuild.contains(indexName) && DatabaseDescriptor.isDaemonInitialized())
SystemKeyspace.setIndexBuilt(baseCfs.keyspace.getName(), indexName);
}
}
}
/**
* Marks the specified index as failed.
* {@link #markIndexesBuilding(Set, boolean, boolean)} should always be invoked before this method.
*
* @param index the index to be marked as built
*/
private synchronized void markIndexFailed(Index index)
{
String indexName = index.getIndexMetadata().name;
AtomicInteger counter = inProgressBuilds.get(indexName);
if (counter != null)
{
assert counter.get() > 0;
counter.decrementAndGet();
if (DatabaseDescriptor.isDaemonInitialized())
SystemKeyspace.setIndexRemoved(baseCfs.keyspace.getName(), indexName);
needsFullRebuild.add(indexName);
}
}
private void logAndMarkIndexesFailed(Set indexes, Throwable indexBuildFailure)
{
JVMStabilityInspector.inspectThrowable(indexBuildFailure);
if (indexBuildFailure != null)
logger.warn("Index build of {} failed. Please run full index rebuild to fix it.", getIndexNames(indexes), indexBuildFailure);
else
logger.warn("Index build of {} failed. Please run full index rebuild to fix it.", getIndexNames(indexes));
indexes.forEach(SecondaryIndexManager.this::markIndexFailed);
}
/**
* Marks the specified index as removed.
*
* @param indexName the index name
*/
private synchronized void markIndexRemoved(String indexName)
{
SystemKeyspace.setIndexRemoved(baseCfs.keyspace.getName(), indexName);
queryableIndexes.remove(indexName);
needsFullRebuild.remove(indexName);
inProgressBuilds.remove(indexName);
}
public Index getIndexByName(String indexName)
{
return indexes.get(indexName);
}
private Index createInstance(IndexMetadata indexDef)
{
Index newIndex;
if (indexDef.isCustom())
{
assert indexDef.options != null;
String className = indexDef.options.get(IndexTarget.CUSTOM_INDEX_OPTION_NAME);
assert !Strings.isNullOrEmpty(className);
try
{
Class indexClass = FBUtilities.classForName(className, "Index");
Constructor ctor = indexClass.getConstructor(ColumnFamilyStore.class, IndexMetadata.class);
newIndex = ctor.newInstance(baseCfs, indexDef);
}
catch (Exception e)
{
throw new RuntimeException(e);
}
}
else
{
newIndex = CassandraIndex.newIndex(baseCfs, indexDef);
}
return newIndex;
}
/**
* Truncate all indexes
*/
public void truncateAllIndexesBlocking(final long truncatedAt)
{
executeAllBlocking(indexes.values().stream(), (index) -> index.getTruncateTask(truncatedAt), null);
}
/**
* Remove all indexes
*/
public void dropAllIndexes()
{
markAllIndexesRemoved();
invalidateAllIndexesBlocking();
}
@VisibleForTesting
public void invalidateAllIndexesBlocking()
{
executeAllBlocking(indexes.values().stream(), Index::getInvalidateTask, null);
}
/**
* Perform a blocking flush all indexes
*/
public void flushAllIndexesBlocking()
{
flushIndexesBlocking(ImmutableSet.copyOf(indexes.values()));
}
/**
* Perform a blocking flush of selected indexes
*/
public void flushIndexesBlocking(Set indexes)
{
flushIndexesBlocking(indexes, null);
}
/**
* Performs a blocking flush of all custom indexes
*/
public void flushAllNonCFSBackedIndexesBlocking()
{
executeAllBlocking(indexes.values()
.stream()
.filter(index -> !index.getBackingTable().isPresent()),
Index::getBlockingFlushTask, null);
}
/**
* Performs a blocking execution of pre-join tasks of all indexes
*/
public void executePreJoinTasksBlocking(boolean hadBootstrap)
{
logger.info("Executing pre-join{} tasks for: {}", hadBootstrap ? " post-bootstrap" : "", this.baseCfs);
executeAllBlocking(indexes.values().stream(), (index) ->
{
return index.getPreJoinTask(hadBootstrap);
}, null);
}
private void flushIndexesBlocking(Set indexes, FutureCallback