Examples with DataStreamsConfig io.cdap.cdap.etl.proto.v2.DataStreamsConfig used on opensource projects

Example 21 with DataStreamsConfig

use of io.cdap.cdap.etl.proto.v2.DataStreamsConfig in project cdap by caskdata.

the class DataStreamsTest method testAlertPublisher.

@Test
public void testAlertPublisher() throws Exception {
    String sinkName = "alertSink";
    final String topic = "alertTopic";
    Schema schema = Schema.recordOf("x", Schema.Field.of("id", Schema.nullableOf(Schema.of(Schema.Type.LONG))));
    StructuredRecord record1 = StructuredRecord.builder(schema).set("id", 1L).build();
    StructuredRecord record2 = StructuredRecord.builder(schema).set("id", 2L).build();
    StructuredRecord alertRecord = StructuredRecord.builder(schema).build();
    /*
     * source --> nullAlert --> sink
     *               |
     *               |--> TMS publisher
     */
    DataStreamsConfig config = DataStreamsConfig.builder().setBatchInterval("5s").addStage(new ETLStage("source", MockSource.getPlugin(schema, ImmutableList.of(record1, record2, alertRecord)))).addStage(new ETLStage("nullAlert", NullAlertTransform.getPlugin("id"))).addStage(new ETLStage("sink", MockSink.getPlugin(sinkName))).addStage(new ETLStage("tms", TMSAlertPublisher.getPlugin(topic, NamespaceId.DEFAULT.getNamespace()))).addConnection("source", "nullAlert").addConnection("nullAlert", "sink").addConnection("nullAlert", "tms").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("AlertTest");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    final Set<StructuredRecord> expectedRecords = ImmutableSet.of(record1, record2);
    final Set<Alert> expectedMessages = ImmutableSet.of(new Alert("nullAlert", new HashMap<String, String>()));
    final DataSetManager<Table> sinkTable = getDataset(sinkName);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            // get alerts from TMS
            try {
                getMessagingAdmin(NamespaceId.DEFAULT.getNamespace()).getTopicProperties(topic);
            } catch (TopicNotFoundException e) {
                return false;
            }
            MessageFetcher messageFetcher = getMessagingContext().getMessageFetcher();
            Set<Alert> actualMessages = new HashSet<>();
            try (CloseableIterator<Message> iter = messageFetcher.fetch(NamespaceId.DEFAULT.getNamespace(), topic, 5, 0)) {
                while (iter.hasNext()) {
                    Message message = iter.next();
                    Alert alert = GSON.fromJson(message.getPayloadAsString(), Alert.class);
                    actualMessages.add(alert);
                }
            }
            // get records from sink
            sinkTable.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sinkTable));
            return expectedRecords.equals(outputRecords) && expectedMessages.equals(actualMessages);
        }
    }, 4, TimeUnit.MINUTES);
    sparkManager.stop();
    sparkManager.waitForStatus(false, 10, 1);
    validateMetric(appId, "source.records.out", 3);
    validateMetric(appId, "nullAlert.records.in", 3);
    validateMetric(appId, "nullAlert.records.out", 2);
    validateMetric(appId, "nullAlert.records.alert", 1);
    validateMetric(appId, "sink.records.in", 2);
    validateMetric(appId, "tms.records.in", 1);
}

Example 22 with DataStreamsConfig

use of io.cdap.cdap.etl.proto.v2.DataStreamsConfig in project cdap by caskdata.

the class DataStreamsTest method testErrorTransform.

@Test
public void testErrorTransform() throws Exception {
    String sink1TableName = "errTestOut1";
    String sink2TableName = "errTestOut2";
    Schema inputSchema = Schema.recordOf("user", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    List<StructuredRecord> input = ImmutableList.of(StructuredRecord.builder(inputSchema).set("name", "Leo").build(), StructuredRecord.builder(inputSchema).set("name", "Ralph").build(), StructuredRecord.builder(inputSchema).set("name", "Don").build(), StructuredRecord.builder(inputSchema).set("name", "Mike").build(), StructuredRecord.builder(inputSchema).set("name", "April").build());
    /*
     *
     * source--> filter1 --> filter2 --> agg1 --> agg2
     *              |           |         |        |
     *              |-----------|---------|--------|--------|--> flatten errors --> sink1
     *                                                      |
     *                                                      |--> filter errors --> sink2
     * arrows coming out the right represent output records
     * arrows coming out the bottom represent error records
     * this will test multiple stages from multiple phases emitting errors to the same stage
     * as well as errors from one stage going to multiple stages
     */
    DataStreamsConfig config = DataStreamsConfig.builder().setBatchInterval("5s").addStage(new ETLStage("source", MockSource.getPlugin(inputSchema, input))).addStage(new ETLStage("filter1", StringValueFilterTransform.getPlugin("name", "Leo"))).addStage(new ETLStage("filter2", StringValueFilterTransform.getPlugin("name", "Ralph"))).addStage(new ETLStage("agg1", GroupFilterAggregator.getPlugin("name", "Don"))).addStage(new ETLStage("agg2", GroupFilterAggregator.getPlugin("name", "Mike"))).addStage(new ETLStage("errorflatten", FlattenErrorTransform.getPlugin())).addStage(new ETLStage("errorfilter", FilterErrorTransform.getPlugin(3))).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1TableName))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2TableName))).addConnection("source", "filter1").addConnection("filter1", "filter2").addConnection("filter2", "agg1").addConnection("agg1", "agg2").addConnection("filter1", "errorflatten").addConnection("filter1", "errorfilter").addConnection("filter2", "errorflatten").addConnection("filter2", "errorfilter").addConnection("agg1", "errorflatten").addConnection("agg1", "errorfilter").addConnection("agg2", "errorflatten").addConnection("agg2", "errorfilter").addConnection("errorflatten", "sink1").addConnection("errorfilter", "sink2").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("ErrTransformTest");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    Schema flattenSchema = Schema.recordOf("erroruser", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("errMsg", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("errCode", Schema.nullableOf(Schema.of(Schema.Type.INT))), Schema.Field.of("errStage", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    final Set<StructuredRecord> expected = ImmutableSet.of(StructuredRecord.builder(flattenSchema).set("name", "Leo").set("errMsg", "bad string value").set("errCode", 1).set("errStage", "filter1").build(), StructuredRecord.builder(flattenSchema).set("name", "Ralph").set("errMsg", "bad string value").set("errCode", 1).set("errStage", "filter2").build(), StructuredRecord.builder(flattenSchema).set("name", "Don").set("errMsg", "bad val").set("errCode", 3).set("errStage", "agg1").build(), StructuredRecord.builder(flattenSchema).set("name", "Mike").set("errMsg", "bad val").set("errCode", 3).set("errStage", "agg2").build());
    final DataSetManager<Table> sink1Table = getDataset(sink1TableName);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink1Table.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sink1Table));
            return expected.equals(outputRecords);
        }
    }, 4, TimeUnit.MINUTES);
    final Set<StructuredRecord> expected2 = ImmutableSet.of(StructuredRecord.builder(inputSchema).set("name", "Leo").build(), StructuredRecord.builder(inputSchema).set("name", "Ralph").build());
    final DataSetManager<Table> sink2Table = getDataset(sink2TableName);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink2Table.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sink2Table));
            return expected2.equals(outputRecords);
        }
    }, 4, TimeUnit.MINUTES);
}

Example 23 with DataStreamsConfig

use of io.cdap.cdap.etl.proto.v2.DataStreamsConfig in project cdap by caskdata.

the class DataStreamsTest method testParallelAggregators.

@Test
public void testParallelAggregators() throws Exception {
    String sink1Name = "pAggOutput1";
    String sink2Name = "pAggOutput2";
    Schema inputSchema = Schema.recordOf("testRecord", Schema.Field.of("user", Schema.of(Schema.Type.STRING)), Schema.Field.of("item", Schema.of(Schema.Type.LONG)));
    List<StructuredRecord> input1 = ImmutableList.of(StructuredRecord.builder(inputSchema).set("user", "samuel").set("item", 1L).build(), StructuredRecord.builder(inputSchema).set("user", "samuel").set("item", 2L).build());
    List<StructuredRecord> input2 = ImmutableList.of(StructuredRecord.builder(inputSchema).set("user", "samuel").set("item", 3L).build(), StructuredRecord.builder(inputSchema).set("user", "john").set("item", 4L).build(), StructuredRecord.builder(inputSchema).set("user", "john").set("item", 3L).build());
    /*
       source1 --|--> agg1 --> sink1
                 |
       source2 --|--> agg2 --> sink2
     */
    DataStreamsConfig pipelineConfig = DataStreamsConfig.builder().setBatchInterval("5s").addStage(new ETLStage("source1", MockSource.getPlugin(inputSchema, input1))).addStage(new ETLStage("source2", MockSource.getPlugin(inputSchema, input2))).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1Name))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2Name))).addStage(new ETLStage("agg1", FieldCountAggregator.getPlugin("user", "string"))).addStage(new ETLStage("agg2", FieldCountAggregator.getPlugin("item", "long"))).addConnection("source1", "agg1").addConnection("source1", "agg2").addConnection("source2", "agg1").addConnection("source2", "agg2").addConnection("agg1", "sink1").addConnection("agg2", "sink2").disableCheckpoints().build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, pipelineConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("ParallelAggApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    Schema outputSchema1 = Schema.recordOf("user.count", Schema.Field.of("user", Schema.of(Schema.Type.STRING)), Schema.Field.of("ct", Schema.of(Schema.Type.LONG)));
    Schema outputSchema2 = Schema.recordOf("item.count", Schema.Field.of("item", Schema.of(Schema.Type.LONG)), Schema.Field.of("ct", Schema.of(Schema.Type.LONG)));
    // check output
    final DataSetManager<Table> sinkManager1 = getDataset(sink1Name);
    final Set<StructuredRecord> expected1 = ImmutableSet.of(StructuredRecord.builder(outputSchema1).set("user", "all").set("ct", 5L).build(), StructuredRecord.builder(outputSchema1).set("user", "samuel").set("ct", 3L).build(), StructuredRecord.builder(outputSchema1).set("user", "john").set("ct", 2L).build());
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sinkManager1.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sinkManager1));
            return expected1.equals(outputRecords);
        }
    }, 1, TimeUnit.MINUTES);
    final DataSetManager<Table> sinkManager2 = getDataset(sink2Name);
    final Set<StructuredRecord> expected2 = ImmutableSet.of(StructuredRecord.builder(outputSchema2).set("item", 0L).set("ct", 5L).build(), StructuredRecord.builder(outputSchema2).set("item", 1L).set("ct", 1L).build(), StructuredRecord.builder(outputSchema2).set("item", 2L).set("ct", 1L).build(), StructuredRecord.builder(outputSchema2).set("item", 3L).set("ct", 2L).build(), StructuredRecord.builder(outputSchema2).set("item", 4L).set("ct", 1L).build());
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sinkManager2.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sinkManager2));
            return expected2.equals(outputRecords);
        }
    }, 1, TimeUnit.MINUTES);
    sparkManager.stop();
    sparkManager.waitForStatus(false, 10, 1);
    validateMetric(appId, "source1.records.out", 2);
    validateMetric(appId, "source2.records.out", 3);
    validateMetric(appId, "agg1.records.in", 5);
    validateMetric(appId, "agg1.records.out", 3);
    validateMetric(appId, "agg2.records.in", 5);
    validateMetric(appId, "agg2.records.out", 5);
    validateMetric(appId, "sink1.records.in", 3);
    validateMetric(appId, "sink2.records.in", 5);
}

Example 24 with DataStreamsConfig

use of io.cdap.cdap.etl.proto.v2.DataStreamsConfig in project cdap by caskdata.

the class DataStreamsTest method testWindower.

@Test
public void testWindower() throws Exception {
    /*
     * source --> window(width=10,interval=1) --> aggregator --> filter --> sink
     */
    Schema schema = Schema.recordOf("data", Schema.Field.of("x", Schema.of(Schema.Type.STRING)));
    List<StructuredRecord> input = ImmutableList.of(StructuredRecord.builder(schema).set("x", "abc").build(), StructuredRecord.builder(schema).set("x", "abc").build(), StructuredRecord.builder(schema).set("x", "abc").build());
    String sinkName = "windowOut";
    // source sleeps 1 second between outputs
    DataStreamsConfig etlConfig = DataStreamsConfig.builder().addStage(new ETLStage("source", MockSource.getPlugin(schema, input, 1000L))).addStage(new ETLStage("window", Window.getPlugin(30, 1))).addStage(new ETLStage("agg", FieldCountAggregator.getPlugin("x", "string"))).addStage(new ETLStage("filter", StringValueFilterTransform.getPlugin("x", "all"))).addStage(new ETLStage("sink", MockSink.getPlugin(sinkName))).addConnection("source", "window").addConnection("window", "agg").addConnection("agg", "filter").addConnection("filter", "sink").setBatchInterval("1s").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("WindowerApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    // the sink should contain at least one record with count of 3, and no records with more than 3.
    // less than 3 if the window doesn't contain all 3 records yet, but there should eventually be a window
    // that contains all 3.
    final DataSetManager<Table> outputManager = getDataset(sinkName);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            outputManager.flush();
            boolean sawThree = false;
            for (StructuredRecord record : MockSink.readOutput(outputManager)) {
                long count = record.get("ct");
                if (count == 3L) {
                    sawThree = true;
                }
                Assert.assertTrue(count <= 3L);
            }
            return sawThree;
        }
    }, 2, TimeUnit.MINUTES);
    sparkManager.stop();
}

Example 25 with DataStreamsConfig

use of io.cdap.cdap.etl.proto.v2.DataStreamsConfig in project cdap by caskdata.

the class PreviewDataStreamsTest method testDataStreamsPreviewRun.

@Test
public void testDataStreamsPreviewRun() throws Exception {
    PreviewManager previewManager = getPreviewManager();
    String sinkTableName = "singleOutput";
    Schema schema = Schema.recordOf("testRecord", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    List<StructuredRecord> records = new ArrayList<>();
    StructuredRecord recordSamuel = StructuredRecord.builder(schema).set("name", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(schema).set("name", "bob").build();
    StructuredRecord recordTest = StructuredRecord.builder(schema).set("name", "test").build();
    records.add(recordSamuel);
    records.add(recordBob);
    records.add(recordTest);
    /*
     * source --> transform -> sink
     */
    DataStreamsConfig etlConfig = DataStreamsConfig.builder().addStage(new ETLStage("source", MockSource.getPlugin(schema, records))).addStage(new ETLStage("transform", IdentityTransform.getPlugin())).addStage(new ETLStage("sink", MockSink.getPlugin(sinkTableName))).addConnection("source", "transform").addConnection("transform", "sink").setNumOfRecordsPreview(100).setBatchInterval("1s").build();
    // Construct the preview config with the program name and program type.
    PreviewConfig previewConfig = new PreviewConfig(DataStreamsSparkLauncher.NAME, ProgramType.SPARK, Collections.<String, String>emptyMap(), 1);
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig, previewConfig);
    // Start the preview and get the corresponding PreviewRunner.
    ApplicationId previewId = previewManager.start(NamespaceId.DEFAULT, appRequest);
    final PreviewRunner previewRunner = previewManager.getRunner(previewId);
    // Wait for the preview to be running and wait until the records are processed in the sink.
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            Map<String, List<JsonElement>> data = previewRunner.getData("sink");
            return data != null && data.get(DATA_TRACER_PROPERTY) != null && data.get(DATA_TRACER_PROPERTY).size() == 3;
        }
    }, 1, TimeUnit.MINUTES);
    // check data in source and transform
    checkPreviewStore(previewRunner, "source", 3);
    checkPreviewStore(previewRunner, "transform", 3);
    // Wait for the pipeline to be shutdown by timer.
    TimeUnit.MINUTES.sleep(1);
    Tasks.waitFor(PreviewStatus.Status.KILLED_BY_TIMER, new Callable<PreviewStatus.Status>() {

        @Override
        public PreviewStatus.Status call() throws Exception {
            return previewRunner.getStatus().getStatus();
        }
    }, 1, TimeUnit.MINUTES);
    // Validate the metrics for preview
    validateMetric(3, previewId, "source.records.out", previewRunner);
    validateMetric(3, previewId, "transform.records.in", previewRunner);
    validateMetric(3, previewId, "transform.records.out", previewRunner);
    validateMetric(3, previewId, "sink.records.in", previewRunner);
    validateMetric(3, previewId, "sink.records.out", previewRunner);
    // Check the sink table is not created in the real space.
    DataSetManager<Table> sinkManager = getDataset(sinkTableName);
    Assert.assertNull(sinkManager.get());
}