/** * Test lineage info is set on publishing multiple tasks */ @Test public void testPublishMultiTasks() throws IOException { WorkUnitState state1 = buildTaskState(2); WorkUnitState state2 = buildTaskState(2); LineageInfo lineageInfo = LineageInfo.getLineageInfo(state1.getTaskBroker()).get(); DatasetDescriptor source = new DatasetDescriptor("kafka", "testTopic"); lineageInfo.setSource(source, state1); lineageInfo.setSource(source, state2); BaseDataPublisher publisher = new BaseDataPublisher(state1); publisher.publishData(ImmutableList.of(state1, state2)); Assert.assertTrue(state1.contains("gobblin.event.lineage.branch.0.destination")); Assert.assertTrue(state1.contains("gobblin.event.lineage.branch.1.destination")); Assert.assertTrue(state2.contains("gobblin.event.lineage.branch.0.destination")); Assert.assertTrue(state2.contains("gobblin.event.lineage.branch.1.destination")); }
/** * Test lineage info is set on publishing single task */ @Test public void testPublishSingleTask() throws IOException { WorkUnitState state = buildTaskState(1); LineageInfo lineageInfo = LineageInfo.getLineageInfo(state.getTaskBroker()).get(); DatasetDescriptor source = new DatasetDescriptor("kafka", "testTopic"); lineageInfo.setSource(source, state); BaseDataPublisher publisher = new BaseDataPublisher(state); publisher.publishData(state); Assert.assertTrue(state.contains("gobblin.event.lineage.branch.0.destination")); Assert.assertFalse(state.contains("gobblin.event.lineage.branch.1.destination")); }
int numPartitionsPerBranch = 2; WorkUnitState state = buildTaskState(numBranches); LineageInfo lineageInfo = LineageInfo.getLineageInfo(state.getTaskBroker()).get(); DatasetDescriptor source = new DatasetDescriptor("kafka", "testTopic");