Examples with SegmentProvider org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider used on opensource projects

Example 11 with SegmentProvider

use of org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider in project druid by druid-io.

the class CompactionTaskTest method testCreateIngestionSchemaWithCustomMetricsSpec.

@Test
public void testCreateIngestionSchemaWithCustomMetricsSpec() throws IOException, SegmentLoadingException {
    final AggregatorFactory[] customMetricsSpec = new AggregatorFactory[] { new CountAggregatorFactory("custom_count"), new LongSumAggregatorFactory("custom_long_sum", "agg_1"), new FloatMinAggregatorFactory("custom_float_min", "agg_3"), new DoubleMaxAggregatorFactory("custom_double_max", "agg_4") };
    final List<ParallelIndexIngestionSpec> ingestionSpecs = CompactionTask.createIngestionSchema(toolbox, LockGranularity.TIME_CHUNK, new SegmentProvider(DATA_SOURCE, new CompactionIntervalSpec(COMPACTION_INTERVAL, null)), new PartitionConfigurationManager(TUNING_CONFIG), null, null, customMetricsSpec, null, COORDINATOR_CLIENT, segmentCacheManagerFactory, RETRY_POLICY_FACTORY, IOConfig.DEFAULT_DROP_EXISTING);
    final List<DimensionsSpec> expectedDimensionsSpec = getExpectedDimensionsSpecForAutoGeneration();
    ingestionSpecs.sort((s1, s2) -> Comparators.intervalsByStartThenEnd().compare(s1.getDataSchema().getGranularitySpec().inputIntervals().get(0), s2.getDataSchema().getGranularitySpec().inputIntervals().get(0)));
    Assert.assertEquals(6, ingestionSpecs.size());
    assertIngestionSchema(ingestionSpecs, expectedDimensionsSpec, Arrays.asList(customMetricsSpec), SEGMENT_INTERVALS, Granularities.MONTH, Granularities.NONE, IOConfig.DEFAULT_DROP_EXISTING);
}

Example 12 with SegmentProvider

use of org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider in project druid by druid-io.

the class CompactionTaskTest method testGranularitySpecWithNullRollup.

@Test
public void testGranularitySpecWithNullRollup() throws IOException, SegmentLoadingException {
    final List<ParallelIndexIngestionSpec> ingestionSpecs = CompactionTask.createIngestionSchema(toolbox, LockGranularity.TIME_CHUNK, new SegmentProvider(DATA_SOURCE, new CompactionIntervalSpec(COMPACTION_INTERVAL, null)), new PartitionConfigurationManager(TUNING_CONFIG), null, null, null, new ClientCompactionTaskGranularitySpec(null, null, null), COORDINATOR_CLIENT, segmentCacheManagerFactory, RETRY_POLICY_FACTORY, IOConfig.DEFAULT_DROP_EXISTING);
    Assert.assertEquals(6, ingestionSpecs.size());
    for (ParallelIndexIngestionSpec indexIngestionSpec : ingestionSpecs) {
        // Expect false since rollup value in metadata of existing segments are null
        Assert.assertFalse(indexIngestionSpec.getDataSchema().getGranularitySpec().isRollup());
    }
}

Example 13 with SegmentProvider

use of org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider in project druid by druid-io.

the class CompactionTaskTest method testCreateIngestionSchemaWithMaxTotalRows.

@Test
public void testCreateIngestionSchemaWithMaxTotalRows() throws IOException, SegmentLoadingException {
    final CompactionTask.CompactionTuningConfig tuningConfig = new CompactionTask.CompactionTuningConfig(null, null, null, 500000, 1000000L, null, 1000000L, null, null, null, new IndexSpec(new RoaringBitmapSerdeFactory(true), CompressionStrategy.LZ4, CompressionStrategy.LZF, LongEncodingStrategy.LONGS), null, null, false, false, 5000L, null, null, null, null, null, null, null, null, null, null, null, null, null, null);
    final List<ParallelIndexIngestionSpec> ingestionSpecs = CompactionTask.createIngestionSchema(toolbox, LockGranularity.TIME_CHUNK, new SegmentProvider(DATA_SOURCE, new CompactionIntervalSpec(COMPACTION_INTERVAL, null)), new PartitionConfigurationManager(tuningConfig), null, null, null, null, COORDINATOR_CLIENT, segmentCacheManagerFactory, RETRY_POLICY_FACTORY, IOConfig.DEFAULT_DROP_EXISTING);
    final List<DimensionsSpec> expectedDimensionsSpec = getExpectedDimensionsSpecForAutoGeneration();
    ingestionSpecs.sort((s1, s2) -> Comparators.intervalsByStartThenEnd().compare(s1.getDataSchema().getGranularitySpec().inputIntervals().get(0), s2.getDataSchema().getGranularitySpec().inputIntervals().get(0)));
    Assert.assertEquals(6, ingestionSpecs.size());
    assertIngestionSchema(ingestionSpecs, expectedDimensionsSpec, AGGREGATORS.stream().map(AggregatorFactory::getCombiningFactory).collect(Collectors.toList()), SEGMENT_INTERVALS, tuningConfig, Granularities.MONTH, Granularities.NONE, IOConfig.DEFAULT_DROP_EXISTING);
}

Example 14 with SegmentProvider

use of org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider in project druid by druid-io.

the class CompactionTaskTest method testSegmentGranularityAndNullQueryGranularity.

@Test
public void testSegmentGranularityAndNullQueryGranularity() throws IOException, SegmentLoadingException {
    final List<ParallelIndexIngestionSpec> ingestionSpecs = CompactionTask.createIngestionSchema(toolbox, LockGranularity.TIME_CHUNK, new SegmentProvider(DATA_SOURCE, new CompactionIntervalSpec(COMPACTION_INTERVAL, null)), new PartitionConfigurationManager(TUNING_CONFIG), null, null, null, new ClientCompactionTaskGranularitySpec(new PeriodGranularity(Period.months(3), null, null), null, null), COORDINATOR_CLIENT, segmentCacheManagerFactory, RETRY_POLICY_FACTORY, IOConfig.DEFAULT_DROP_EXISTING);
    final List<DimensionsSpec> expectedDimensionsSpec = ImmutableList.of(new DimensionsSpec(getDimensionSchema(new DoubleDimensionSchema("string_to_double"))));
    ingestionSpecs.sort((s1, s2) -> Comparators.intervalsByStartThenEnd().compare(s1.getDataSchema().getGranularitySpec().inputIntervals().get(0), s2.getDataSchema().getGranularitySpec().inputIntervals().get(0)));
    Assert.assertEquals(1, ingestionSpecs.size());
    assertIngestionSchema(ingestionSpecs, expectedDimensionsSpec, AGGREGATORS.stream().map(AggregatorFactory::getCombiningFactory).collect(Collectors.toList()), Collections.singletonList(COMPACTION_INTERVAL), new PeriodGranularity(Period.months(3), null, null), Granularities.NONE, IOConfig.DEFAULT_DROP_EXISTING);
}

Example 15 with SegmentProvider

use of org.apache.druid.indexing.common.task.CompactionTask.SegmentProvider in project druid by druid-io.

the class CompactionTaskTest method testQueryGranularityAndNullSegmentGranularity.

@Test
public void testQueryGranularityAndNullSegmentGranularity() throws IOException, SegmentLoadingException {
    final List<ParallelIndexIngestionSpec> ingestionSpecs = CompactionTask.createIngestionSchema(toolbox, LockGranularity.TIME_CHUNK, new SegmentProvider(DATA_SOURCE, new CompactionIntervalSpec(COMPACTION_INTERVAL, null)), new PartitionConfigurationManager(TUNING_CONFIG), null, null, null, new ClientCompactionTaskGranularitySpec(null, new PeriodGranularity(Period.months(3), null, null), null), COORDINATOR_CLIENT, segmentCacheManagerFactory, RETRY_POLICY_FACTORY, IOConfig.DEFAULT_DROP_EXISTING);
    final List<DimensionsSpec> expectedDimensionsSpec = getExpectedDimensionsSpecForAutoGeneration();
    ingestionSpecs.sort((s1, s2) -> Comparators.intervalsByStartThenEnd().compare(s1.getDataSchema().getGranularitySpec().inputIntervals().get(0), s2.getDataSchema().getGranularitySpec().inputIntervals().get(0)));
    Assert.assertEquals(6, ingestionSpecs.size());
    assertIngestionSchema(ingestionSpecs, expectedDimensionsSpec, AGGREGATORS.stream().map(AggregatorFactory::getCombiningFactory).collect(Collectors.toList()), SEGMENT_INTERVALS, Granularities.MONTH, new PeriodGranularity(Period.months(3), null, null), IOConfig.DEFAULT_DROP_EXISTING);
}