Examples with BasicLineageWriter io.cdap.cdap.data2.metadata.writer.BasicLineageWriter used on opensource projects

Example 1 with BasicLineageWriter

use of io.cdap.cdap.data2.metadata.writer.BasicLineageWriter in project cdap by caskdata.

the class LineageAdminTest method testSimpleLineage.

@Test
public void testSimpleLineage() {
    // Lineage for D3 -> P2 -> D2 -> P1 -> D1
    TransactionRunner transactionRunner = getInjector().getInstance(TransactionRunner.class);
    LineageStoreReader lineageReader = new DefaultLineageStoreReader(transactionRunner);
    LineageWriter lineageWriter = new BasicLineageWriter(transactionRunner);
    Store store = getInjector().getInstance(Store.class);
    LineageAdmin lineageAdmin = new LineageAdmin(lineageReader, store);
    // Add accesses for D3 -> P2 -> D2 -> P1 -> D1 <-> P3
    // We need to use current time here as metadata store stores access time using current time
    ProgramRunId run1 = program1.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId run2 = program2.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId run3 = program3.run(RunIds.generate(System.currentTimeMillis()).getId());
    addRuns(store, run1, run2, run3);
    // It is okay to use current time here since access time is ignore during assertions
    lineageWriter.addAccess(run1, dataset1, AccessType.UNKNOWN);
    lineageWriter.addAccess(run1, dataset1, AccessType.WRITE);
    lineageWriter.addAccess(run1, dataset2, AccessType.READ);
    lineageWriter.addAccess(run2, dataset2, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset3, AccessType.READ);
    lineageWriter.addAccess(run3, dataset1, AccessType.UNKNOWN, null);
    // The UNKNOWN access type will get filtered out if there is READ/WRITE. It will be preserved if it is the
    // only access type
    Lineage expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, program1, AccessType.READ, twillRunId(run1)), new Relation(dataset2, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset3, program2, AccessType.READ, twillRunId(run2)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3))));
    // Lineage for D1
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 100));
    // Lineage for D2
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset2, 500, System.currentTimeMillis() + 10000, 100));
    // Lineage for D1 for one level should be D2 -> P1 -> D1 <-> P3
    Lineage oneLevelLineage = lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 1);
    Assert.assertEquals(ImmutableSet.of(new Relation(dataset1, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, program1, AccessType.READ, twillRunId(run1)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3))), oneLevelLineage.getRelations());
    // Assert that in a different namespace both lineage and metadata should be empty
    NamespaceId customNamespace = new NamespaceId("custom_namespace");
    DatasetId customDataset1 = customNamespace.dataset(dataset1.getEntityName());
    Assert.assertEquals(new Lineage(ImmutableSet.of()), lineageAdmin.computeLineage(customDataset1, 500, System.currentTimeMillis() + 10000, 100));
}

Example 2 with BasicLineageWriter

use of io.cdap.cdap.data2.metadata.writer.BasicLineageWriter in project cdap by caskdata.

the class LineageAdminTest method testDirectCycle.

@Test
public void testDirectCycle() {
    // Lineage for:
    // 
    // D1 <-> P1
    // 
    TransactionRunner transactionRunner = getInjector().getInstance(TransactionRunner.class);
    LineageStoreReader lineageReader = new DefaultLineageStoreReader(transactionRunner);
    LineageWriter lineageWriter = new BasicLineageWriter(transactionRunner);
    Store store = getInjector().getInstance(Store.class);
    LineageAdmin lineageAdmin = new LineageAdmin(lineageReader, store);
    // Add accesses
    addRuns(store, run1, run2, run3, run4, run5);
    // It is okay to use current time here since access time is ignore during assertions
    lineageWriter.addAccess(run1, dataset1, AccessType.READ);
    lineageWriter.addAccess(run1, dataset1, AccessType.WRITE);
    Lineage expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset1, program1, AccessType.READ, twillRunId(run1))));
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset1, 500, 20000, 100));
}

Example 3 with BasicLineageWriter

use of io.cdap.cdap.data2.metadata.writer.BasicLineageWriter in project cdap by caskdata.

the class LineageAdminTest method testWorkflowLineage.

@Test
public void testWorkflowLineage() {
    TransactionRunner transactionRunner = getInjector().getInstance(TransactionRunner.class);
    LineageStoreReader lineageReader = new DefaultLineageStoreReader(transactionRunner);
    LineageWriter lineageWriter = new BasicLineageWriter(transactionRunner);
    ApplicationId testApp = NamespaceId.DEFAULT.app("testApp");
    ProgramId workflowId = testApp.workflow("wf1");
    // if the spark and mr job are inner jobs of workflow, they should be in the same app
    ProgramId mrId = testApp.mr("mr1");
    ProgramId sparkId = testApp.mr("spark1");
    ImmutableList<WorkflowNode> nodes = ImmutableList.of(new WorkflowActionNode("mr1", new ScheduleProgramInfo(SchedulableProgramType.MAPREDUCE, "mr1")), new WorkflowActionNode("spark1", new ScheduleProgramInfo(SchedulableProgramType.SPARK, "spark1")));
    WorkflowSpecification wfSpec = new WorkflowSpecification("test", "wf1", "", Collections.emptyMap(), nodes, Collections.emptyMap(), Collections.emptyMap());
    ApplicationSpecification appSpec = new DefaultApplicationSpecification("testApp", ProjectInfo.getVersion().toString(), "dummy app", null, NamespaceId.DEFAULT.artifact("testArtifact", "1.0").toApiArtifactId(), Collections.emptyMap(), Collections.emptyMap(), Collections.emptyMap(), Collections.emptyMap(), ImmutableMap.of(workflowId.getProgram(), wfSpec), Collections.emptyMap(), Collections.emptyMap(), Collections.emptyMap(), Collections.emptyMap());
    Store store = getInjector().getInstance(Store.class);
    store.addApplication(testApp, appSpec);
    LineageAdmin lineageAdmin = new LineageAdmin(lineageReader, store);
    // Add accesses for D3 -> P2 -> D2 -> P1 -> D1 <-> P3
    // |
    // |-> P5,
    // P1 and P2 are inner programs of the workflow
    // We need to use current time here as metadata store stores access time using current time
    ProgramRunId run1 = mrId.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId run2 = sparkId.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId run3 = program3.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId workflow = workflowId.run(RunIds.generate(System.currentTimeMillis()).getId());
    ProgramRunId run5 = program5.run(RunIds.generate(System.currentTimeMillis()).getId());
    addRuns(store, workflow);
    // only mr and spark can be inner programs
    addWorkflowRuns(store, workflow.getProgram(), workflow.getRun(), run1, run2);
    addRuns(store, run3);
    addRuns(store, run5);
    // It is okay to use current time here since access time is ignore during assertions
    lineageWriter.addAccess(run1, dataset1, AccessType.WRITE);
    lineageWriter.addAccess(run1, dataset2, AccessType.READ);
    lineageWriter.addAccess(run2, dataset2, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset3, AccessType.READ);
    lineageWriter.addAccess(run3, dataset1, AccessType.UNKNOWN, null);
    lineageWriter.addAccess(run5, dataset1, AccessType.READ, null);
    // The UNKNOWN access type will get filtered out if there is READ/WRITE. It will be preserved if it is the
    // only access type
    Lineage expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, workflowId, AccessType.WRITE, twillRunId(workflow)), new Relation(dataset2, workflowId, AccessType.READ, twillRunId(workflow)), new Relation(dataset2, workflowId, AccessType.WRITE, twillRunId(workflow)), new Relation(dataset3, workflowId, AccessType.READ, twillRunId(workflow)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3)), new Relation(dataset1, program5, AccessType.READ, twillRunId(run5))));
    Lineage resultLineage = lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 100, "workflow");
    // Lineage for D1
    Assert.assertEquals(expectedLineage, resultLineage);
    resultLineage = lineageAdmin.computeLineage(dataset2, 500, System.currentTimeMillis() + 10000, 100, "workflow");
    // Lineage for D2
    Assert.assertEquals(expectedLineage, resultLineage);
    // Lineage for D1 for one level should be D2 -> P1 -> D1 <-> P3
    Lineage oneLevelLineage = lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 1, "workflow");
    Assert.assertEquals(ImmutableSet.of(new Relation(dataset1, workflowId, AccessType.WRITE, twillRunId(workflow)), new Relation(dataset2, workflowId, AccessType.READ, twillRunId(workflow)), new Relation(dataset1, program5, AccessType.READ, twillRunId(run5)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3))), oneLevelLineage.getRelations());
    // Run tests without workflow parameter
    expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, mrId, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, mrId, AccessType.READ, twillRunId(run1)), new Relation(dataset2, sparkId, AccessType.WRITE, twillRunId(run2)), new Relation(dataset3, sparkId, AccessType.READ, twillRunId(run2)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3)), new Relation(dataset1, program5, AccessType.READ, twillRunId(run5))));
    resultLineage = lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 100, null);
    // Lineage for D1
    Assert.assertEquals(expectedLineage, resultLineage);
    resultLineage = lineageAdmin.computeLineage(dataset2, 500, System.currentTimeMillis() + 10000, 100, null);
    // Lineage for D2
    Assert.assertEquals(expectedLineage, resultLineage);
    // Lineage for D1 for one level should be D2 -> P1 -> D1 <-> P3
    oneLevelLineage = lineageAdmin.computeLineage(dataset1, 500, System.currentTimeMillis() + 10000, 1, null);
    Assert.assertEquals(ImmutableSet.of(new Relation(dataset1, mrId, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, mrId, AccessType.READ, twillRunId(run1)), new Relation(dataset1, program5, AccessType.READ, twillRunId(run5)), new Relation(dataset1, program3, AccessType.UNKNOWN, twillRunId(run3))), oneLevelLineage.getRelations());
    // Assert that in a different namespace both lineage and metadata should be empty
    NamespaceId customNamespace = new NamespaceId("custom_namespace");
    DatasetId customDataset1 = customNamespace.dataset(dataset1.getEntityName());
    Assert.assertEquals(new Lineage(ImmutableSet.of()), lineageAdmin.computeLineage(customDataset1, 500, System.currentTimeMillis() + 10000, 100));
}

Example 4 with BasicLineageWriter

use of io.cdap.cdap.data2.metadata.writer.BasicLineageWriter in project cdap by caskdata.

the class LineageAdminTest method testBranchLoopLineage.

@Test
public void testBranchLoopLineage() {
    // Lineage for:
    // 
    // |-------------------------------------|
    // |                                     |
    // |                                     |
    // |    -> D4       -> D5 -> P3 -> D6 -> P5
    // |    |           |                    ^
    // V    |           |                    |
    // D1 -> P1 -> D2 -> P2 -> D3 ----------->|
    // |     |           |
    // |     |           |
    // S1 -->|     ---------------> P4 -> D7
    TransactionRunner transactionRunner = getInjector().getInstance(TransactionRunner.class);
    LineageStoreReader lineageReader = new DefaultLineageStoreReader(transactionRunner);
    LineageWriter lineageWriter = new BasicLineageWriter(transactionRunner);
    Store store = getInjector().getInstance(Store.class);
    LineageAdmin lineageAdmin = new LineageAdmin(lineageReader, store);
    // Add accesses
    addRuns(store, run1, run2, run3, run4, run5);
    // It is okay to use current time here since access time is ignore during assertions
    lineageWriter.addAccess(run1, dataset1, AccessType.READ);
    lineageWriter.addAccess(run1, dataset2, AccessType.WRITE);
    lineageWriter.addAccess(run1, dataset4, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset2, AccessType.READ);
    lineageWriter.addAccess(run2, dataset3, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset5, AccessType.WRITE);
    lineageWriter.addAccess(run3, dataset5, AccessType.READ, null);
    lineageWriter.addAccess(run3, dataset6, AccessType.WRITE, null);
    lineageWriter.addAccess(run4, dataset2, AccessType.READ, null);
    lineageWriter.addAccess(run4, dataset3, AccessType.READ, null);
    lineageWriter.addAccess(run4, dataset7, AccessType.WRITE, null);
    lineageWriter.addAccess(run5, dataset3, AccessType.READ, null);
    lineageWriter.addAccess(run5, dataset6, AccessType.READ, null);
    lineageWriter.addAccess(run5, dataset1, AccessType.WRITE, null);
    Lineage expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, program1, AccessType.READ, twillRunId(run1)), new Relation(dataset2, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset4, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, program2, AccessType.READ, twillRunId(run2)), new Relation(dataset3, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program3, AccessType.READ, twillRunId(run3)), new Relation(dataset6, program3, AccessType.WRITE, twillRunId(run3)), new Relation(dataset2, program4, AccessType.READ, twillRunId(run4)), new Relation(dataset3, program4, AccessType.READ, twillRunId(run4)), new Relation(dataset7, program4, AccessType.WRITE, twillRunId(run4)), new Relation(dataset3, program5, AccessType.READ, twillRunId(run5)), new Relation(dataset6, program5, AccessType.READ, twillRunId(run5)), new Relation(dataset1, program5, AccessType.WRITE, twillRunId(run5))));
    // Lineage for D1
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset1, 500, 20000, 100));
    // Lineage for D5
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset5, 500, 20000, 100));
    // Lineage for D7
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset7, 500, 20000, 100));
    // Lineage for D5 for one level
    // -> D5 -> P3 -> D6
    // |
    // |
    // D2 -> P2 -> D3
    Lineage oneLevelLineage = lineageAdmin.computeLineage(dataset5, 500, 20000, 1);
    Assert.assertEquals(ImmutableSet.of(new Relation(dataset2, program2, AccessType.READ, twillRunId(run2)), new Relation(dataset3, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program3, AccessType.READ, twillRunId(run3)), new Relation(dataset6, program3, AccessType.WRITE, twillRunId(run3))), oneLevelLineage.getRelations());
}

Example 5 with BasicLineageWriter

use of io.cdap.cdap.data2.metadata.writer.BasicLineageWriter in project cdap by caskdata.

the class LineageAdminTest method testBranchLineage.

@Test
public void testBranchLineage() {
    // Lineage for:
    // 
    // ->D4        -> D5 -> P3 -> D6
    // |           |
    // |           |
    // D1 -> P1 -> D2 -> P2 -> D3
    // |     |           |
    // |     |           |
    // S1 -->|     ---------------> P4 -> D7
    TransactionRunner transactionRunner = getInjector().getInstance(TransactionRunner.class);
    LineageStoreReader lineageReader = new DefaultLineageStoreReader(transactionRunner);
    LineageWriter lineageWriter = new BasicLineageWriter(transactionRunner);
    Store store = getInjector().getInstance(Store.class);
    LineageAdmin lineageAdmin = new LineageAdmin(lineageReader, store);
    // Add accesses
    addRuns(store, run1, run2, run3, run4, run5);
    // It is okay to use current time here since access time is ignore during assertions
    lineageWriter.addAccess(run1, dataset1, AccessType.READ);
    lineageWriter.addAccess(run1, dataset2, AccessType.WRITE);
    lineageWriter.addAccess(run1, dataset4, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset2, AccessType.READ);
    lineageWriter.addAccess(run2, dataset3, AccessType.WRITE);
    lineageWriter.addAccess(run2, dataset5, AccessType.WRITE);
    lineageWriter.addAccess(run3, dataset5, AccessType.READ, null);
    lineageWriter.addAccess(run3, dataset6, AccessType.WRITE, null);
    lineageWriter.addAccess(run4, dataset2, AccessType.READ, null);
    lineageWriter.addAccess(run4, dataset3, AccessType.READ, null);
    lineageWriter.addAccess(run4, dataset7, AccessType.WRITE, null);
    Lineage expectedLineage = new Lineage(ImmutableSet.of(new Relation(dataset1, program1, AccessType.READ, twillRunId(run1)), new Relation(dataset2, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset4, program1, AccessType.WRITE, twillRunId(run1)), new Relation(dataset2, program2, AccessType.READ, twillRunId(run2)), new Relation(dataset3, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program2, AccessType.WRITE, twillRunId(run2)), new Relation(dataset5, program3, AccessType.READ, twillRunId(run3)), new Relation(dataset6, program3, AccessType.WRITE, twillRunId(run3)), new Relation(dataset2, program4, AccessType.READ, twillRunId(run4)), new Relation(dataset3, program4, AccessType.READ, twillRunId(run4)), new Relation(dataset7, program4, AccessType.WRITE, twillRunId(run4))));
    // Lineage for D7
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset7, 500, 20000, 100));
    // Lineage for D6
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset6, 500, 20000, 100));
    // Lineage for D3
    Assert.assertEquals(expectedLineage, lineageAdmin.computeLineage(dataset3, 500, 20000, 100));
}