@Override
public Set<DataSegment> findUsedSegments(Set<SegmentIdentifier> identifiers)
throws IOException {
final VersionedIntervalTimeline<String, DataSegment> timeline =
new VersionedIntervalTimeline<>(Ordering.natural());
for (DataSegment dataSegment : appenderatorTester.getPushedSegments()) {
timeline.add(
dataSegment.getInterval(),
dataSegment.getVersion(),
dataSegment.getShardSpec().createChunk(dataSegment));
}
final Set<DataSegment> retVal = Sets.newHashSet();
for (SegmentIdentifier identifier : identifiers) {
for (TimelineObjectHolder<String, DataSegment> holder :
timeline.lookup(identifier.getInterval())) {
for (PartitionChunk<DataSegment> chunk : holder.getObject()) {
if (identifiers.contains(SegmentIdentifier.fromDataSegment(chunk.getObject()))) {
retVal.add(chunk.getObject());
}
}
}
}
return retVal;
}
@Override
public Sequence<T> run(final Query<T> query, final Map<String, Object> responseContext) {
final QueryToolChest<T, Query<T>> toolChest = warehouse.getToolChest(query);
final CacheStrategy<T, Object, Query<T>> strategy = toolChest.getCacheStrategy(query);
final Map<DruidServer, List<SegmentDescriptor>> serverSegments = Maps.newTreeMap();
final List<Pair<Interval, byte[]>> cachedResults = Lists.newArrayList();
final Map<String, CachePopulator> cachePopulatorMap = Maps.newHashMap();
final boolean useCache =
BaseQuery.getContextUseCache(query, true)
&& strategy != null
&& cacheConfig.isUseCache()
&& cacheConfig.isQueryCacheable(query);
final boolean populateCache =
BaseQuery.getContextPopulateCache(query, true)
&& strategy != null
&& cacheConfig.isPopulateCache()
&& cacheConfig.isQueryCacheable(query);
final boolean isBySegment = BaseQuery.getContextBySegment(query, false);
final ImmutableMap.Builder<String, Object> contextBuilder = new ImmutableMap.Builder<>();
final int priority = BaseQuery.getContextPriority(query, 0);
contextBuilder.put("priority", priority);
if (populateCache) {
// prevent down-stream nodes from caching results as well if we are populating the cache
contextBuilder.put(CacheConfig.POPULATE_CACHE, false);
contextBuilder.put("bySegment", true);
}
TimelineLookup<String, ServerSelector> timeline = serverView.getTimeline(query.getDataSource());
if (timeline == null) {
return Sequences.empty();
}
// build set of segments to query
Set<Pair<ServerSelector, SegmentDescriptor>> segments = Sets.newLinkedHashSet();
List<TimelineObjectHolder<String, ServerSelector>> serversLookup = Lists.newLinkedList();
// Note that enabling this leads to putting uncovered intervals information in the response
// headers
// and might blow up in some cases https://github.com/druid-io/druid/issues/2108
int uncoveredIntervalsLimit = BaseQuery.getContextUncoveredIntervalsLimit(query, 0);
if (uncoveredIntervalsLimit > 0) {
List<Interval> uncoveredIntervals = Lists.newArrayListWithCapacity(uncoveredIntervalsLimit);
boolean uncoveredIntervalsOverflowed = false;
for (Interval interval : query.getIntervals()) {
Iterable<TimelineObjectHolder<String, ServerSelector>> lookup = timeline.lookup(interval);
long startMillis = interval.getStartMillis();
long endMillis = interval.getEndMillis();
for (TimelineObjectHolder<String, ServerSelector> holder : lookup) {
Interval holderInterval = holder.getInterval();
long intervalStart = holderInterval.getStartMillis();
if (!uncoveredIntervalsOverflowed && startMillis != intervalStart) {
if (uncoveredIntervalsLimit > uncoveredIntervals.size()) {
uncoveredIntervals.add(new Interval(startMillis, intervalStart));
} else {
uncoveredIntervalsOverflowed = true;
}
}
startMillis = holderInterval.getEndMillis();
serversLookup.add(holder);
}
if (!uncoveredIntervalsOverflowed && startMillis < endMillis) {
if (uncoveredIntervalsLimit > uncoveredIntervals.size()) {
uncoveredIntervals.add(new Interval(startMillis, endMillis));
} else {
uncoveredIntervalsOverflowed = true;
}
}
}
if (!uncoveredIntervals.isEmpty()) {
// This returns intervals for which NO segment is present.
// Which is not necessarily an indication that the data doesn't exist or is
// incomplete. The data could exist and just not be loaded yet. In either
// case, though, this query will not include any data from the identified intervals.
responseContext.put("uncoveredIntervals", uncoveredIntervals);
responseContext.put("uncoveredIntervalsOverflowed", uncoveredIntervalsOverflowed);
}
} else {
for (Interval interval : query.getIntervals()) {
Iterables.addAll(serversLookup, timeline.lookup(interval));
}
}
// Let tool chest filter out unneeded segments
final List<TimelineObjectHolder<String, ServerSelector>> filteredServersLookup =
toolChest.filterSegments(query, serversLookup);
Map<String, Optional<RangeSet<String>>> dimensionRangeCache = Maps.newHashMap();
// Filter unneeded chunks based on partition dimension
for (TimelineObjectHolder<String, ServerSelector> holder : filteredServersLookup) {
final Set<PartitionChunk<ServerSelector>> filteredChunks =
DimFilterUtils.filterShards(
query.getFilter(),
holder.getObject(),
new Function<PartitionChunk<ServerSelector>, ShardSpec>() {
@Override
public ShardSpec apply(PartitionChunk<ServerSelector> input) {
return input.getObject().getSegment().getShardSpec();
}
},
dimensionRangeCache);
for (PartitionChunk<ServerSelector> chunk : filteredChunks) {
ServerSelector selector = chunk.getObject();
final SegmentDescriptor descriptor =
new SegmentDescriptor(
holder.getInterval(), holder.getVersion(), chunk.getChunkNumber());
segments.add(Pair.of(selector, descriptor));
}
}
final byte[] queryCacheKey;
if ((populateCache || useCache) // implies strategy != null
&& !isBySegment) // explicit bySegment queries are never cached
{
queryCacheKey = strategy.computeCacheKey(query);
} else {
queryCacheKey = null;
}
if (queryCacheKey != null) {
// cachKeys map must preserve segment ordering, in order for shards to always be combined in
// the same order
Map<Pair<ServerSelector, SegmentDescriptor>, Cache.NamedKey> cacheKeys =
Maps.newLinkedHashMap();
for (Pair<ServerSelector, SegmentDescriptor> segment : segments) {
final Cache.NamedKey segmentCacheKey =
CacheUtil.computeSegmentCacheKey(
segment.lhs.getSegment().getIdentifier(), segment.rhs, queryCacheKey);
cacheKeys.put(segment, segmentCacheKey);
}
// Pull cached segments from cache and remove from set of segments to query
final Map<Cache.NamedKey, byte[]> cachedValues;
if (useCache) {
cachedValues =
cache.getBulk(
Iterables.limit(cacheKeys.values(), cacheConfig.getCacheBulkMergeLimit()));
} else {
cachedValues = ImmutableMap.of();
}
for (Map.Entry<Pair<ServerSelector, SegmentDescriptor>, Cache.NamedKey> entry :
cacheKeys.entrySet()) {
Pair<ServerSelector, SegmentDescriptor> segment = entry.getKey();
Cache.NamedKey segmentCacheKey = entry.getValue();
final Interval segmentQueryInterval = segment.rhs.getInterval();
final byte[] cachedValue = cachedValues.get(segmentCacheKey);
if (cachedValue != null) {
// remove cached segment from set of segments to query
segments.remove(segment);
cachedResults.add(Pair.of(segmentQueryInterval, cachedValue));
} else if (populateCache) {
// otherwise, if populating cache, add segment to list of segments to cache
final String segmentIdentifier = segment.lhs.getSegment().getIdentifier();
cachePopulatorMap.put(
String.format("%s_%s", segmentIdentifier, segmentQueryInterval),
new CachePopulator(cache, objectMapper, segmentCacheKey));
}
}
}
// Compile list of all segments not pulled from cache
for (Pair<ServerSelector, SegmentDescriptor> segment : segments) {
final QueryableDruidServer queryableDruidServer = segment.lhs.pick();
if (queryableDruidServer == null) {
log.makeAlert(
"No servers found for SegmentDescriptor[%s] for DataSource[%s]?! How can this be?!",
segment.rhs, query.getDataSource())
.emit();
} else {
final DruidServer server = queryableDruidServer.getServer();
List<SegmentDescriptor> descriptors = serverSegments.get(server);
if (descriptors == null) {
descriptors = Lists.newArrayList();
serverSegments.put(server, descriptors);
}
descriptors.add(segment.rhs);
}
}
return new LazySequence<>(
new Supplier<Sequence<T>>() {
@Override
public Sequence<T> get() {
ArrayList<Sequence<T>> sequencesByInterval = Lists.newArrayList();
addSequencesFromCache(sequencesByInterval);
addSequencesFromServer(sequencesByInterval);
return mergeCachedAndUncachedSequences(query, sequencesByInterval);
}
private void addSequencesFromCache(ArrayList<Sequence<T>> listOfSequences) {
if (strategy == null) {
return;
}
final Function<Object, T> pullFromCacheFunction = strategy.pullFromCache();
final TypeReference<Object> cacheObjectClazz = strategy.getCacheObjectClazz();
for (Pair<Interval, byte[]> cachedResultPair : cachedResults) {
final byte[] cachedResult = cachedResultPair.rhs;
Sequence<Object> cachedSequence =
new BaseSequence<>(
new BaseSequence.IteratorMaker<Object, Iterator<Object>>() {
@Override
public Iterator<Object> make() {
try {
if (cachedResult.length == 0) {
return Iterators.emptyIterator();
}
return objectMapper.readValues(
objectMapper.getFactory().createParser(cachedResult),
cacheObjectClazz);
} catch (IOException e) {
throw Throwables.propagate(e);
}
}
@Override
public void cleanup(Iterator<Object> iterFromMake) {}
});
listOfSequences.add(Sequences.map(cachedSequence, pullFromCacheFunction));
}
}
private void addSequencesFromServer(ArrayList<Sequence<T>> listOfSequences) {
listOfSequences.ensureCapacity(listOfSequences.size() + serverSegments.size());
final Query<T> rewrittenQuery = query.withOverriddenContext(contextBuilder.build());
// Loop through each server, setting up the query and initiating it.
// The data gets handled as a Future and parsed in the long Sequence chain in the
// resultSeqToAdd setter.
for (Map.Entry<DruidServer, List<SegmentDescriptor>> entry :
serverSegments.entrySet()) {
final DruidServer server = entry.getKey();
final List<SegmentDescriptor> descriptors = entry.getValue();
final QueryRunner clientQueryable = serverView.getQueryRunner(server);
if (clientQueryable == null) {
log.error("WTF!? server[%s] doesn't have a client Queryable?", server);
continue;
}
final MultipleSpecificSegmentSpec segmentSpec =
new MultipleSpecificSegmentSpec(descriptors);
final Sequence<T> resultSeqToAdd;
if (!server.isAssignable()
|| !populateCache
|| isBySegment) { // Direct server queryable
if (!isBySegment) {
resultSeqToAdd =
clientQueryable.run(query.withQuerySegmentSpec(segmentSpec), responseContext);
} else {
// bySegment queries need to be de-serialized, see DirectDruidClient.run()
@SuppressWarnings("unchecked")
final Query<Result<BySegmentResultValueClass<T>>> bySegmentQuery =
(Query<Result<BySegmentResultValueClass<T>>>) ((Query) query);
@SuppressWarnings("unchecked")
final Sequence<Result<BySegmentResultValueClass<T>>> resultSequence =
clientQueryable.run(
bySegmentQuery.withQuerySegmentSpec(segmentSpec), responseContext);
resultSeqToAdd =
(Sequence)
Sequences.map(
resultSequence,
new Function<
Result<BySegmentResultValueClass<T>>,
Result<BySegmentResultValueClass<T>>>() {
@Override
public Result<BySegmentResultValueClass<T>> apply(
Result<BySegmentResultValueClass<T>> input) {
final BySegmentResultValueClass<T> bySegmentValue =
input.getValue();
return new Result<>(
input.getTimestamp(),
new BySegmentResultValueClass<T>(
Lists.transform(
bySegmentValue.getResults(),
toolChest.makePreComputeManipulatorFn(
query, MetricManipulatorFns.deserializing())),
bySegmentValue.getSegmentId(),
bySegmentValue.getInterval()));
}
});
}
} else { // Requires some manipulation on broker side
@SuppressWarnings("unchecked")
final Sequence<Result<BySegmentResultValueClass<T>>> runningSequence =
clientQueryable.run(
rewrittenQuery.withQuerySegmentSpec(segmentSpec), responseContext);
resultSeqToAdd =
new MergeSequence(
query.getResultOrdering(),
Sequences.<Result<BySegmentResultValueClass<T>>, Sequence<T>>map(
runningSequence,
new Function<Result<BySegmentResultValueClass<T>>, Sequence<T>>() {
private final Function<T, Object> cacheFn =
strategy.prepareForCache();
// Acctually do something with the results
@Override
public Sequence<T> apply(Result<BySegmentResultValueClass<T>> input) {
final BySegmentResultValueClass<T> value = input.getValue();
final CachePopulator cachePopulator =
cachePopulatorMap.get(
String.format(
"%s_%s", value.getSegmentId(), value.getInterval()));
final Queue<ListenableFuture<Object>> cacheFutures =
new ConcurrentLinkedQueue<>();
return Sequences.<T>withEffect(
Sequences.<T, T>map(
Sequences.<T, T>map(
Sequences.<T>simple(value.getResults()),
new Function<T, T>() {
@Override
public T apply(final T input) {
if (cachePopulator != null) {
// only compute cache data if populating cache
cacheFutures.add(
backgroundExecutorService.submit(
new Callable<Object>() {
@Override
public Object call() {
return cacheFn.apply(input);
}
}));
}
return input;
}
}),
toolChest.makePreComputeManipulatorFn(
// Ick... most makePreComputeManipulatorFn directly cast
// to their ToolChest query type of choice
// This casting is sub-optimal, but hasn't caused any
// major problems yet...
(Query) rewrittenQuery,
MetricManipulatorFns.deserializing())),
new Runnable() {
@Override
public void run() {
if (cachePopulator != null) {
Futures.addCallback(
Futures.allAsList(cacheFutures),
new FutureCallback<List<Object>>() {
@Override
public void onSuccess(List<Object> cacheData) {
cachePopulator.populate(cacheData);
// Help out GC by making sure all references are
// gone
cacheFutures.clear();
}
@Override
public void onFailure(Throwable throwable) {
log.error(throwable, "Background caching failed");
}
},
backgroundExecutorService);
}
}
},
MoreExecutors.sameThreadExecutor()); // End withEffect
}
}));
}
listOfSequences.add(resultSeqToAdd);
}
}
} // End of Supplier
);
}
@Override
public Sequence<T> run(final Query<T> query) {
final QueryToolChest<T, Query<T>> toolChest = warehouse.getToolChest(query);
final CacheStrategy<T, Object, Query<T>> strategy = toolChest.getCacheStrategy(query);
final Map<DruidServer, List<SegmentDescriptor>> serverSegments = Maps.newTreeMap();
final List<Pair<DateTime, byte[]>> cachedResults = Lists.newArrayList();
final Map<String, CachePopulator> cachePopulatorMap = Maps.newHashMap();
final boolean useCache =
Boolean.parseBoolean(query.getContextValue("useCache", "true")) && strategy != null;
final boolean populateCache =
Boolean.parseBoolean(query.getContextValue("populateCache", "true")) && strategy != null;
final boolean isBySegment = Boolean.parseBoolean(query.getContextValue("bySegment", "false"));
ImmutableMap.Builder<String, String> contextBuilder = new ImmutableMap.Builder<>();
final String priority = query.getContextValue("priority", "0");
contextBuilder.put("priority", priority);
if (populateCache) {
contextBuilder.put("bySegment", "true");
}
contextBuilder.put("intermediate", "true");
final Query<T> rewrittenQuery = query.withOverriddenContext(contextBuilder.build());
VersionedIntervalTimeline<String, ServerSelector> timeline =
serverView.getTimeline(query.getDataSource());
if (timeline == null) {
return Sequences.empty();
}
// build set of segments to query
Set<Pair<ServerSelector, SegmentDescriptor>> segments = Sets.newLinkedHashSet();
List<TimelineObjectHolder<String, ServerSelector>> serversLookup = Lists.newLinkedList();
for (Interval interval : rewrittenQuery.getIntervals()) {
serversLookup.addAll(timeline.lookup(interval));
}
// Let tool chest filter out unneeded segments
final List<TimelineObjectHolder<String, ServerSelector>> filteredServersLookup =
toolChest.filterSegments(query, serversLookup);
for (TimelineObjectHolder<String, ServerSelector> holder : filteredServersLookup) {
for (PartitionChunk<ServerSelector> chunk : holder.getObject()) {
ServerSelector selector = chunk.getObject();
final SegmentDescriptor descriptor =
new SegmentDescriptor(
holder.getInterval(), holder.getVersion(), chunk.getChunkNumber());
segments.add(Pair.of(selector, descriptor));
}
}
final byte[] queryCacheKey;
if (strategy != null) {
queryCacheKey = strategy.computeCacheKey(query);
} else {
queryCacheKey = null;
}
if (queryCacheKey != null) {
Map<Pair<ServerSelector, SegmentDescriptor>, Cache.NamedKey> cacheKeys = Maps.newHashMap();
for (Pair<ServerSelector, SegmentDescriptor> segment : segments) {
final Cache.NamedKey segmentCacheKey =
computeSegmentCacheKey(
segment.lhs.getSegment().getIdentifier(), segment.rhs, queryCacheKey);
cacheKeys.put(segment, segmentCacheKey);
}
// Pull cached segments from cache and remove from set of segments to query
final Map<Cache.NamedKey, byte[]> cachedValues;
if (useCache) {
cachedValues = cache.getBulk(cacheKeys.values());
} else {
cachedValues = ImmutableMap.of();
}
for (Map.Entry<Pair<ServerSelector, SegmentDescriptor>, Cache.NamedKey> entry :
cacheKeys.entrySet()) {
Pair<ServerSelector, SegmentDescriptor> segment = entry.getKey();
Cache.NamedKey segmentCacheKey = entry.getValue();
final Interval segmentQueryInterval = segment.rhs.getInterval();
final byte[] cachedValue = cachedValues.get(segmentCacheKey);
if (cachedValue != null) {
// remove cached segment from set of segments to query
segments.remove(segment);
cachedResults.add(Pair.of(segmentQueryInterval.getStart(), cachedValue));
} else if (populateCache) {
final String segmentIdentifier = segment.lhs.getSegment().getIdentifier();
cachePopulatorMap.put(
String.format("%s_%s", segmentIdentifier, segmentQueryInterval),
new CachePopulator(cache, objectMapper, segmentCacheKey));
}
}
}
// Compile list of all segments not pulled from cache
for (Pair<ServerSelector, SegmentDescriptor> segment : segments) {
final QueryableDruidServer queryableDruidServer = segment.lhs.pick();
if (queryableDruidServer == null) {
log.error("No servers found for %s?! How can this be?!", segment.rhs);
} else {
final DruidServer server = queryableDruidServer.getServer();
List<SegmentDescriptor> descriptors = serverSegments.get(server);
if (descriptors == null) {
descriptors = Lists.newArrayList();
serverSegments.put(server, descriptors);
}
descriptors.add(segment.rhs);
}
}
return new LazySequence<>(
new Supplier<Sequence<T>>() {
@Override
public Sequence<T> get() {
ArrayList<Pair<DateTime, Sequence<T>>> listOfSequences = Lists.newArrayList();
addSequencesFromServer(listOfSequences);
addSequencesFromCache(listOfSequences);
Collections.sort(
listOfSequences,
Ordering.natural().onResultOf(Pair.<DateTime, Sequence<T>>lhsFn()));
final Sequence<Sequence<T>> seq =
Sequences.simple(
Iterables.transform(listOfSequences, Pair.<DateTime, Sequence<T>>rhsFn()));
if (strategy == null) {
return toolChest.mergeSequences(seq);
} else {
return strategy.mergeSequences(seq);
}
}
private void addSequencesFromCache(
ArrayList<Pair<DateTime, Sequence<T>>> listOfSequences) {
if (strategy == null) {
return;
}
final Function<Object, T> pullFromCacheFunction = strategy.pullFromCache();
final TypeReference<Object> cacheObjectClazz = strategy.getCacheObjectClazz();
for (Pair<DateTime, byte[]> cachedResultPair : cachedResults) {
final byte[] cachedResult = cachedResultPair.rhs;
Sequence<Object> cachedSequence =
new BaseSequence<>(
new BaseSequence.IteratorMaker<Object, Iterator<Object>>() {
@Override
public Iterator<Object> make() {
try {
if (cachedResult.length == 0) {
return Iterators.emptyIterator();
}
return objectMapper.readValues(
objectMapper.getFactory().createParser(cachedResult),
cacheObjectClazz);
} catch (IOException e) {
throw Throwables.propagate(e);
}
}
@Override
public void cleanup(Iterator<Object> iterFromMake) {}
});
listOfSequences.add(
Pair.of(
cachedResultPair.lhs, Sequences.map(cachedSequence, pullFromCacheFunction)));
}
}
@SuppressWarnings("unchecked")
private void addSequencesFromServer(
ArrayList<Pair<DateTime, Sequence<T>>> listOfSequences) {
for (Map.Entry<DruidServer, List<SegmentDescriptor>> entry :
serverSegments.entrySet()) {
final DruidServer server = entry.getKey();
final List<SegmentDescriptor> descriptors = entry.getValue();
final QueryRunner clientQueryable = serverView.getQueryRunner(server);
if (clientQueryable == null) {
log.makeAlert("WTF!? server[%s] doesn't have a client Queryable?", server).emit();
continue;
}
final Sequence<T> resultSeqToAdd;
final MultipleSpecificSegmentSpec segmentSpec =
new MultipleSpecificSegmentSpec(descriptors);
List<Interval> intervals = segmentSpec.getIntervals();
if (!server.isAssignable() || !populateCache || isBySegment) {
resultSeqToAdd = clientQueryable.run(query.withQuerySegmentSpec(segmentSpec));
} else {
resultSeqToAdd =
toolChest.mergeSequences(
Sequences.map(
clientQueryable.run(rewrittenQuery.withQuerySegmentSpec(segmentSpec)),
new Function<Object, Sequence<T>>() {
private final Function<T, Object> prepareForCache =
strategy.prepareForCache();
@Override
public Sequence<T> apply(Object input) {
Result<Object> result = (Result<Object>) input;
final BySegmentResultValueClass<T> value =
(BySegmentResultValueClass<T>) result.getValue();
String segmentIdentifier = value.getSegmentId();
final Iterable<T> segmentResults = value.getResults();
CachePopulator cachePopulator =
cachePopulatorMap.get(
String.format(
"%s_%s", segmentIdentifier, value.getInterval()));
if (cachePopulator != null) {
cachePopulator.populate(
Iterables.transform(segmentResults, prepareForCache));
}
return Sequences.simple(
Iterables.transform(
segmentResults,
toolChest.makeMetricManipulatorFn(
rewrittenQuery,
new MetricManipulationFn() {
@Override
public Object manipulate(
AggregatorFactory factory, Object object) {
return factory.deserialize(object);
}
})));
}
}));
}
listOfSequences.add(Pair.of(intervals.get(0).getStart(), resultSeqToAdd));
}
}
});
}