Examples with LinearShardSpec org.apache.druid.timeline.partition.LinearShardSpec used on opensource projects

Example 1 with LinearShardSpec

use of org.apache.druid.timeline.partition.LinearShardSpec in project druid by druid-io.

the class SqlVectorizedExpressionSanityTest method setupClass.

@BeforeClass
public static void setupClass() {
    Calcites.setSystemProperties();
    ExpressionProcessing.initializeForStrictBooleansTests(true);
    CLOSER = Closer.create();
    final GeneratorSchemaInfo schemaInfo = GeneratorBasicSchemas.SCHEMA_MAP.get("expression-testbench");
    final DataSegment dataSegment = DataSegment.builder().dataSource("foo").interval(schemaInfo.getDataInterval()).version("1").shardSpec(new LinearShardSpec(0)).size(0).build();
    final SegmentGenerator segmentGenerator = CLOSER.register(new SegmentGenerator());
    INDEX = CLOSER.register(segmentGenerator.generate(dataSegment, schemaInfo, Granularities.HOUR, ROWS_PER_SEGMENT));
    CONGLOMERATE = QueryStackTests.createQueryRunnerFactoryConglomerate(CLOSER);
    WALKER = new SpecificSegmentsQuerySegmentWalker(CONGLOMERATE).add(dataSegment, INDEX);
    CLOSER.register(WALKER);
    final PlannerConfig plannerConfig = new PlannerConfig();
    final DruidSchemaCatalog rootSchema = CalciteTests.createMockRootSchema(CONGLOMERATE, WALKER, plannerConfig, AuthTestUtils.TEST_AUTHORIZER_MAPPER);
    PLANNER_FACTORY = new PlannerFactory(rootSchema, CalciteTests.createMockQueryMakerFactory(WALKER, CONGLOMERATE), CalciteTests.createOperatorTable(), CalciteTests.createExprMacroTable(), plannerConfig, AuthTestUtils.TEST_AUTHORIZER_MAPPER, CalciteTests.getJsonMapper(), CalciteTests.DRUID_SCHEMA_NAME);
}

Example 2 with LinearShardSpec

use of org.apache.druid.timeline.partition.LinearShardSpec in project druid by druid-io.

the class IndexerSQLMetadataStorageCoordinatorTest method testAllocatePendingSegmentAfterDroppingExistingSegment.

/**
 * This test simulates an issue detected on the field consisting of the following sequence of events:
 * - A kafka stream segment was created on a given interval
 * - Later, after the above was published, another segment on same interval was created by the stream
 * - Later, after the above was published, another segment on same interval was created by the stream
 * - Later a compaction was issued for the three segments above
 * - Later, after the above was published, another segment on same interval was created by the stream
 * - Later, the compacted segment got dropped due to a drop rule
 * - Later, after the above was dropped, another segment on same interval was created by the stream but this
 *   time there was an integrity violation in the pending segments table because the
 *   {@link IndexerSQLMetadataStorageCoordinator#createNewSegment(Handle, String, Interval, PartialShardSpec, String)}
 *   method returned an segment id that already existed in the pending segments table
 */
@Test
public void testAllocatePendingSegmentAfterDroppingExistingSegment() {
    String maxVersion = "version_newer_newer";
    // simulate one load using kafka streaming
    final PartialShardSpec partialShardSpec = NumberedPartialShardSpec.instance();
    final String dataSource = "ds";
    final Interval interval = Intervals.of("2017-01-01/2017-02-01");
    final SegmentIdWithShardSpec identifier = coordinator.allocatePendingSegment(dataSource, "seq", null, interval, partialShardSpec, "version", true);
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version", identifier.toString());
    // simulate one more load using kafka streaming (as if previous segment was published, note different sequence name)
    final SegmentIdWithShardSpec identifier1 = coordinator.allocatePendingSegment(dataSource, "seq2", identifier.toString(), interval, partialShardSpec, maxVersion, true);
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version_1", identifier1.toString());
    // simulate one more load using kafka streaming (as if previous segment was published, note different sequence name)
    final SegmentIdWithShardSpec identifier2 = coordinator.allocatePendingSegment(dataSource, "seq3", identifier1.toString(), interval, partialShardSpec, maxVersion, true);
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version_2", identifier2.toString());
    // now simulate that one compaction was done (batch) ingestion for same interval (like reindex of the previous three):
    DataSegment segment = new DataSegment("ds", Intervals.of("2017-01-01T00Z/2017-02-01T00Z"), "version_new", ImmutableMap.of(), ImmutableList.of("dim1"), ImmutableList.of("m1"), new LinearShardSpec(0), 9, 100);
    Assert.assertTrue(insertUsedSegments(ImmutableSet.of(segment)));
    List<String> ids = retrieveUsedSegmentIds();
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version_new", ids.get(0));
    // one more load on same interval:
    final SegmentIdWithShardSpec identifier3 = coordinator.allocatePendingSegment(dataSource, "seq4", identifier1.toString(), interval, partialShardSpec, maxVersion, true);
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version_new_1", identifier3.toString());
    // now drop the used segment previously loaded:
    markAllSegmentsUnused(ImmutableSet.of(segment));
    // and final load, this reproduces an issue that could happen with multiple streaming appends,
    // followed by a reindex, followed by a drop, and more streaming data coming in for same interval
    final SegmentIdWithShardSpec identifier4 = coordinator.allocatePendingSegment(dataSource, "seq5", identifier1.toString(), interval, partialShardSpec, maxVersion, true);
    Assert.assertEquals("ds_2017-01-01T00:00:00.000Z_2017-02-01T00:00:00.000Z_version_new_2", identifier4.toString());
}

Example 3 with LinearShardSpec

use of org.apache.druid.timeline.partition.LinearShardSpec in project druid by druid-io.

the class IndexerSQLMetadataStorageCoordinatorTest method testAnnounceHistoricalSegments.

@Test
public void testAnnounceHistoricalSegments() throws IOException {
    Set<DataSegment> segments = new HashSet<>();
    for (int i = 0; i < 105; i++) {
        segments.add(new DataSegment("fooDataSource", Intervals.of("2015-01-01T00Z/2015-01-02T00Z"), "version", ImmutableMap.of(), ImmutableList.of("dim1"), ImmutableList.of("m1"), new LinearShardSpec(i), 9, 100));
    }
    coordinator.announceHistoricalSegments(segments);
    for (DataSegment segment : segments) {
        Assert.assertArrayEquals(mapper.writeValueAsString(segment).getBytes(StandardCharsets.UTF_8), derbyConnector.lookup(derbyConnectorRule.metadataTablesConfigSupplier().get().getSegmentsTable(), "id", "payload", segment.getId().toString()));
    }
    List<String> segmentIds = segments.stream().map(segment -> segment.getId().toString()).collect(Collectors.toList());
    segmentIds.sort(Comparator.naturalOrder());
    Assert.assertEquals(segmentIds, retrieveUsedSegmentIds());
    // Should not update dataSource metadata.
    Assert.assertEquals(0, metadataUpdateCounter.get());
}

Example 4 with LinearShardSpec

use of org.apache.druid.timeline.partition.LinearShardSpec in project druid by druid-io.

the class DruidSchemaTest method setUp.

@Before
public void setUp() throws Exception {
    final File tmpDir = temporaryFolder.newFolder();
    final QueryableIndex index1 = IndexBuilder.create().tmpDir(new File(tmpDir, "1")).segmentWriteOutMediumFactory(OffHeapMemorySegmentWriteOutMediumFactory.instance()).schema(new IncrementalIndexSchema.Builder().withMetrics(new CountAggregatorFactory("cnt"), new DoubleSumAggregatorFactory("m1", "m1"), new HyperUniquesAggregatorFactory("unique_dim1", "dim1")).withRollup(false).build()).rows(ROWS1).buildMMappedIndex();
    final QueryableIndex index2 = IndexBuilder.create().tmpDir(new File(tmpDir, "2")).segmentWriteOutMediumFactory(OffHeapMemorySegmentWriteOutMediumFactory.instance()).schema(new IncrementalIndexSchema.Builder().withMetrics(new LongSumAggregatorFactory("m1", "m1")).withRollup(false).build()).rows(ROWS2).buildMMappedIndex();
    walker = new SpecificSegmentsQuerySegmentWalker(conglomerate).add(DataSegment.builder().dataSource(CalciteTests.DATASOURCE1).interval(Intervals.of("2000/P1Y")).version("1").shardSpec(new LinearShardSpec(0)).size(0).build(), index1).add(DataSegment.builder().dataSource(CalciteTests.DATASOURCE1).interval(Intervals.of("2001/P1Y")).version("1").shardSpec(new LinearShardSpec(0)).size(0).build(), index2).add(DataSegment.builder().dataSource(CalciteTests.DATASOURCE2).interval(index2.getDataInterval()).version("1").shardSpec(new LinearShardSpec(0)).size(0).build(), index2);
    final DataSegment segment1 = new DataSegment("foo3", Intervals.of("2012/2013"), "version3", null, ImmutableList.of("dim1", "dim2"), ImmutableList.of("met1", "met2"), new NumberedShardSpec(2, 3), null, 1, 100L, PruneSpecsHolder.DEFAULT);
    final List<DataSegment> realtimeSegments = ImmutableList.of(segment1);
    serverView = new TestServerInventoryView(walker.getSegments(), realtimeSegments);
    druidServers = serverView.getDruidServers();
    schema = new DruidSchema(CalciteTests.createMockQueryLifecycleFactory(walker, conglomerate), serverView, segmentManager, new MapJoinableFactory(ImmutableSet.of(globalTableJoinable), ImmutableMap.of(globalTableJoinable.getClass(), GlobalTableDataSource.class)), PLANNER_CONFIG_DEFAULT, new NoopEscalator(), new BrokerInternalQueryConfig(), null) {

        @Override
        protected DruidTable buildDruidTable(String dataSource) {
            DruidTable table = super.buildDruidTable(dataSource);
            buildTableLatch.countDown();
            return table;
        }

        @Override
        void markDataSourceAsNeedRebuild(String datasource) {
            super.markDataSourceAsNeedRebuild(datasource);
            markDataSourceLatch.countDown();
        }
    };
    schema2 = new DruidSchema(CalciteTests.createMockQueryLifecycleFactory(walker, conglomerate), serverView, segmentManager, new MapJoinableFactory(ImmutableSet.of(globalTableJoinable), ImmutableMap.of(globalTableJoinable.getClass(), GlobalTableDataSource.class)), PLANNER_CONFIG_DEFAULT, new NoopEscalator(), new BrokerInternalQueryConfig(), null) {

        boolean throwException = true;

        @Override
        protected DruidTable buildDruidTable(String dataSource) {
            DruidTable table = super.buildDruidTable(dataSource);
            buildTableLatch.countDown();
            return table;
        }

        @Override
        protected Set<SegmentId> refreshSegments(final Set<SegmentId> segments) throws IOException {
            if (throwException) {
                throwException = false;
                throw new RuntimeException("Query[xxxx] url[http://xxxx:8083/druid/v2/] timed out.");
            } else {
                return super.refreshSegments(segments);
            }
        }

        @Override
        void markDataSourceAsNeedRebuild(String datasource) {
            super.markDataSourceAsNeedRebuild(datasource);
            markDataSourceLatch.countDown();
        }
    };
    schema.start();
    schema.awaitInitialization();
}

Example 5 with LinearShardSpec

use of org.apache.druid.timeline.partition.LinearShardSpec in project druid by druid-io.

the class ExpressionVectorSelectorsTest method setupClass.

@BeforeClass
public static void setupClass() {
    CLOSER = Closer.create();
    final GeneratorSchemaInfo schemaInfo = GeneratorBasicSchemas.SCHEMA_MAP.get("expression-testbench");
    final DataSegment dataSegment = DataSegment.builder().dataSource("foo").interval(schemaInfo.getDataInterval()).version("1").shardSpec(new LinearShardSpec(0)).size(0).build();
    final SegmentGenerator segmentGenerator = CLOSER.register(new SegmentGenerator());
    INDEX = CLOSER.register(segmentGenerator.generate(dataSegment, schemaInfo, Granularities.HOUR, ROWS_PER_SEGMENT));
}