Examples with Segment org.apache.druid.segment.Segment used on opensource projects

Example 11 with Segment

use of org.apache.druid.segment.Segment in project druid by druid-io.

the class IncrementalIndexTest method testSingleThreadedIndexingAndQuery.

@Test
public void testSingleThreadedIndexingAndQuery() throws Exception {
    final int dimensionCount = 5;
    final ArrayList<AggregatorFactory> ingestAggregatorFactories = new ArrayList<>();
    ingestAggregatorFactories.add(new CountAggregatorFactory("rows"));
    for (int i = 0; i < dimensionCount; ++i) {
        ingestAggregatorFactories.add(new LongSumAggregatorFactory(StringUtils.format("sumResult%s", i), StringUtils.format("Dim_%s", i)));
        ingestAggregatorFactories.add(new DoubleSumAggregatorFactory(StringUtils.format("doubleSumResult%s", i), StringUtils.format("Dim_%s", i)));
    }
    final IncrementalIndex index = indexCreator.createIndex((Object) ingestAggregatorFactories.toArray(new AggregatorFactory[0]));
    final long timestamp = System.currentTimeMillis();
    final int rows = 50;
    // ingesting same data twice to have some merging happening
    for (int i = 0; i < rows; i++) {
        index.add(getLongRow(timestamp + i, dimensionCount));
    }
    for (int i = 0; i < rows; i++) {
        index.add(getLongRow(timestamp + i, dimensionCount));
    }
    // run a timeseries query on the index and verify results
    final ArrayList<AggregatorFactory> queryAggregatorFactories = new ArrayList<>();
    queryAggregatorFactories.add(new CountAggregatorFactory("rows"));
    for (int i = 0; i < dimensionCount; ++i) {
        queryAggregatorFactories.add(new LongSumAggregatorFactory(StringUtils.format("sumResult%s", i), StringUtils.format("sumResult%s", i)));
        queryAggregatorFactories.add(new DoubleSumAggregatorFactory(StringUtils.format("doubleSumResult%s", i), StringUtils.format("doubleSumResult%s", i)));
    }
    TimeseriesQuery query = Druids.newTimeseriesQueryBuilder().dataSource("xxx").granularity(Granularities.ALL).intervals(ImmutableList.of(Intervals.of("2000/2030"))).aggregators(queryAggregatorFactories).build();
    final Segment incrementalIndexSegment = new IncrementalIndexSegment(index, null);
    final QueryRunnerFactory factory = new TimeseriesQueryRunnerFactory(new TimeseriesQueryQueryToolChest(), new TimeseriesQueryEngine(), QueryRunnerTestHelper.NOOP_QUERYWATCHER);
    final QueryRunner<Result<TimeseriesResultValue>> runner = new FinalizeResultsQueryRunner<Result<TimeseriesResultValue>>(factory.createRunner(incrementalIndexSegment), factory.getToolchest());
    List<Result<TimeseriesResultValue>> results = runner.run(QueryPlus.wrap(query)).toList();
    Result<TimeseriesResultValue> result = Iterables.getOnlyElement(results);
    boolean isRollup = index.isRollup();
    Assert.assertEquals(rows * (isRollup ? 1 : 2), result.getValue().getLongMetric("rows").intValue());
    for (int i = 0; i < dimensionCount; ++i) {
        Assert.assertEquals("Failed long sum on dimension " + i, 2 * rows, result.getValue().getLongMetric("sumResult" + i).intValue());
        Assert.assertEquals("Failed double sum on dimension " + i, 2 * rows, result.getValue().getDoubleMetric("doubleSumResult" + i).intValue());
    }
}

Example 12 with Segment

use of org.apache.druid.segment.Segment in project druid by druid-io.

the class SegmentManagerTest method assertResult.

// TODO remove when the bug in intelliJ is fixed.
@SuppressWarnings("RedundantThrows")
private void assertResult(List<DataSegment> expectedExistingSegments) throws SegmentLoadingException {
    final Map<String, Long> expectedDataSourceSizes = expectedExistingSegments.stream().collect(Collectors.toMap(DataSegment::getDataSource, DataSegment::getSize, Long::sum));
    final Map<String, Long> expectedDataSourceCounts = expectedExistingSegments.stream().collect(Collectors.toMap(DataSegment::getDataSource, segment -> 1L, Long::sum));
    final Set<String> expectedDataSourceNames = expectedExistingSegments.stream().map(DataSegment::getDataSource).collect(Collectors.toSet());
    final Map<String, VersionedIntervalTimeline<String, ReferenceCountingSegment>> expectedTimelines = new HashMap<>();
    for (DataSegment segment : expectedExistingSegments) {
        final VersionedIntervalTimeline<String, ReferenceCountingSegment> expectedTimeline = expectedTimelines.computeIfAbsent(segment.getDataSource(), k -> new VersionedIntervalTimeline<>(Ordering.natural()));
        expectedTimeline.add(segment.getInterval(), segment.getVersion(), segment.getShardSpec().createChunk(ReferenceCountingSegment.wrapSegment(SEGMENT_LOADER.getSegment(segment, false, SegmentLazyLoadFailCallback.NOOP), segment.getShardSpec())));
    }
    Assert.assertEquals(expectedDataSourceNames, segmentManager.getDataSourceNames());
    Assert.assertEquals(expectedDataSourceCounts, segmentManager.getDataSourceCounts());
    Assert.assertEquals(expectedDataSourceSizes, segmentManager.getDataSourceSizes());
    final Map<String, DataSourceState> dataSources = segmentManager.getDataSources();
    Assert.assertEquals(expectedTimelines.size(), dataSources.size());
    dataSources.forEach((sourceName, dataSourceState) -> {
        Assert.assertEquals(expectedDataSourceCounts.get(sourceName).longValue(), dataSourceState.getNumSegments());
        Assert.assertEquals(expectedDataSourceSizes.get(sourceName).longValue(), dataSourceState.getTotalSegmentSize());
        Assert.assertEquals(expectedTimelines.get(sourceName).getAllTimelineEntries(), dataSourceState.getTimeline().getAllTimelineEntries());
    });
}

Example 13 with Segment

use of org.apache.druid.segment.Segment in project druid by druid-io.

the class SegmentManagerThreadSafetyTest method testLoadSameSegment.

@Test(timeout = 6000L)
public void testLoadSameSegment() throws IOException, ExecutionException, InterruptedException {
    final DataSegment segment = createSegment("2019-01-01/2019-01-02");
    final List<Future> futures = IntStream.range(0, 16).mapToObj(i -> exec.submit(() -> segmentManager.loadSegment(segment, false, SegmentLazyLoadFailCallback.NOOP))).collect(Collectors.toList());
    for (Future future : futures) {
        future.get();
    }
    Assert.assertEquals(1, segmentPuller.numFileLoaded.size());
    Assert.assertEquals(1, segmentPuller.numFileLoaded.values().iterator().next().intValue());
    Assert.assertEquals(0, segmentCacheManager.getSegmentLocks().size());
}

Example 14 with Segment

use of org.apache.druid.segment.Segment in project druid by druid-io.

the class SegmentManagerThreadSafetyTest method testLoadMultipleSegments.

@Test(timeout = 6000L)
public void testLoadMultipleSegments() throws IOException, ExecutionException, InterruptedException {
    final List<DataSegment> segments = new ArrayList<>(88);
    for (int i = 0; i < 11; i++) {
        for (int j = 0; j < 8; j++) {
            segments.add(createSegment(StringUtils.format("2019-%02d-01/2019-%02d-01", i + 1, i + 2)));
        }
    }
    final List<Future> futures = IntStream.range(0, 16).mapToObj(i -> exec.submit(() -> {
        for (DataSegment segment : segments) {
            try {
                segmentManager.loadSegment(segment, false, SegmentLazyLoadFailCallback.NOOP);
            } catch (SegmentLoadingException e) {
                throw new RuntimeException(e);
            }
        }
    })).collect(Collectors.toList());
    for (Future future : futures) {
        future.get();
    }
    Assert.assertEquals(11, segmentPuller.numFileLoaded.size());
    Assert.assertEquals(1, segmentPuller.numFileLoaded.values().iterator().next().intValue());
    Assert.assertEquals(0, segmentCacheManager.getSegmentLocks().size());
}