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

Example 1 with DataStreamsConfig

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

the class DataStreamsApp method configure.

@Override
public void configure() {
    DataStreamsConfig config = getConfig();
    setDescription(Objects.firstNonNull(config.getDescription(), "Data Streams Application"));
    DataStreamsPipelineSpec spec = new DataStreamsPipelineSpecGenerator<>(getConfigurer(), ImmutableSet.of(StreamingSource.PLUGIN_TYPE), ImmutableSet.of(BatchSink.PLUGIN_TYPE, SparkSink.PLUGIN_TYPE, AlertPublisher.PLUGIN_TYPE)).generateSpec(config);
    addSpark(new DataStreamsSparkLauncher(spec));
    if (!config.checkpointsDisabled()) {
        createDataset(CHECKPOINT_FILESET, FileSet.class);
    }
}

Example 2 with DataStreamsConfig

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

the class DataStreamsSparkSinkTest method testSparkSink.

@Test
public // stream-rate-updater thread in Spark.
void testSparkSink() throws Exception {
    Schema schema = Schema.recordOf("test", Schema.Field.of("id", Schema.of(Schema.Type.STRING)), Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    List<StructuredRecord> input = new ArrayList<>();
    StructuredRecord samuelRecord = StructuredRecord.builder(schema).set("id", "0").set("name", "samuel").build();
    StructuredRecord jacksonRecord = StructuredRecord.builder(schema).set("id", "1").set("name", "jackson").build();
    StructuredRecord dwayneRecord = StructuredRecord.builder(schema).set("id", "2").set("name", "dwayne").build();
    StructuredRecord johnsonRecord = StructuredRecord.builder(schema).set("id", "3").set("name", "johnson").build();
    input.add(samuelRecord);
    input.add(jacksonRecord);
    input.add(dwayneRecord);
    input.add(johnsonRecord);
    DataStreamsConfig etlConfig = DataStreamsConfig.builder().addStage(new ETLStage("source", MockSource.getPlugin(schema, input))).addStage(new ETLStage("sink", co.cask.cdap.etl.mock.spark.streaming.MockSink.getPlugin("${tablename}"))).addConnection("source", "sink").setBatchInterval("1s").build();
    ApplicationId appId = NamespaceId.DEFAULT.app("sparksinkapp");
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    testSparkSink(appManager, "output1");
    testSparkSink(appManager, "output2");
}

Example 3 with DataStreamsConfig

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

the class DataStreamsTest method testAggregatorJoinerMacrosWithCheckpoints.

@Test
public void testAggregatorJoinerMacrosWithCheckpoints() throws Exception {
    /*
                 |--> aggregator --> sink1
        users1 --|
                 |----|
                      |--> dupeFlagger --> sink2
        users2 -------|
     */
    Schema userSchema = Schema.recordOf("user", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    List<StructuredRecord> users1 = ImmutableList.of(StructuredRecord.builder(userSchema).set("id", 1L).set("name", "Samuel").build(), StructuredRecord.builder(userSchema).set("id", 2L).set("name", "Dwayne").build(), StructuredRecord.builder(userSchema).set("id", 3L).set("name", "Terry").build());
    List<StructuredRecord> users2 = ImmutableList.of(StructuredRecord.builder(userSchema).set("id", 1L).set("name", "Samuel").build(), StructuredRecord.builder(userSchema).set("id", 2L).set("name", "Dwayne").build(), StructuredRecord.builder(userSchema).set("id", 4L).set("name", "Terry").build(), StructuredRecord.builder(userSchema).set("id", 5L).set("name", "Christopher").build());
    DataStreamsConfig pipelineConfig = DataStreamsConfig.builder().setBatchInterval("5s").addStage(new ETLStage("users1", MockSource.getPlugin(userSchema, users1))).addStage(new ETLStage("users2", MockSource.getPlugin(userSchema, users2))).addStage(new ETLStage("sink1", MockSink.getPlugin("sink1"))).addStage(new ETLStage("sink2", MockSink.getPlugin("sink2"))).addStage(new ETLStage("aggregator", FieldCountAggregator.getPlugin("${aggfield}", "${aggType}"))).addStage(new ETLStage("dupeFlagger", DupeFlagger.getPlugin("users1", "${flagField}"))).addConnection("users1", "aggregator").addConnection("aggregator", "sink1").addConnection("users1", "dupeFlagger").addConnection("users2", "dupeFlagger").addConnection("dupeFlagger", "sink2").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, pipelineConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("ParallelAggApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // run it once with this set of macros
    Map<String, String> arguments = new HashMap<>();
    arguments.put("aggfield", "id");
    arguments.put("aggType", "long");
    arguments.put("flagField", "isDupe");
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start(arguments);
    sparkManager.waitForStatus(true, 10, 1);
    final DataSetManager<Table> sink1 = getDataset("sink1");
    final DataSetManager<Table> sink2 = getDataset("sink2");
    Schema aggSchema = Schema.recordOf("user.count", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("ct", Schema.of(Schema.Type.LONG)));
    final Set<StructuredRecord> expectedAggregates = ImmutableSet.of(StructuredRecord.builder(aggSchema).set("id", 0L).set("ct", 3L).build(), StructuredRecord.builder(aggSchema).set("id", 1L).set("ct", 1L).build(), StructuredRecord.builder(aggSchema).set("id", 2L).set("ct", 1L).build(), StructuredRecord.builder(aggSchema).set("id", 3L).set("ct", 1L).build());
    Schema outputSchema = Schema.recordOf("user.flagged", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("isDupe", Schema.of(Schema.Type.BOOLEAN)));
    final Set<StructuredRecord> expectedJoined = ImmutableSet.of(StructuredRecord.builder(outputSchema).set("id", 1L).set("name", "Samuel").set("isDupe", true).build(), StructuredRecord.builder(outputSchema).set("id", 2L).set("name", "Dwayne").set("isDupe", true).build(), StructuredRecord.builder(outputSchema).set("id", 3L).set("name", "Terry").set("isDupe", false).build());
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink1.flush();
            sink2.flush();
            Set<StructuredRecord> actualAggs = new HashSet<>();
            Set<StructuredRecord> actualJoined = new HashSet<>();
            actualAggs.addAll(MockSink.readOutput(sink1));
            actualJoined.addAll(MockSink.readOutput(sink2));
            return expectedAggregates.equals(actualAggs) && expectedJoined.equals(actualJoined);
        }
    }, 1, TimeUnit.MINUTES);
    sparkManager.stop();
    sparkManager.waitForStatus(false, 30, 1);
    MockSink.clear(sink1);
    MockSink.clear(sink2);
    // run it again with different macros to make sure they are re-evaluated and not stored in the checkpoint
    arguments = new HashMap<>();
    arguments.put("aggfield", "name");
    arguments.put("aggType", "string");
    arguments.put("flagField", "dupe");
    sparkManager.start(arguments);
    sparkManager.waitForStatus(true, 10, 1);
    aggSchema = Schema.recordOf("user.count", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("ct", Schema.of(Schema.Type.LONG)));
    final Set<StructuredRecord> expectedAggregates2 = ImmutableSet.of(StructuredRecord.builder(aggSchema).set("name", "all").set("ct", 3L).build(), StructuredRecord.builder(aggSchema).set("name", "Samuel").set("ct", 1L).build(), StructuredRecord.builder(aggSchema).set("name", "Dwayne").set("ct", 1L).build(), StructuredRecord.builder(aggSchema).set("name", "Terry").set("ct", 1L).build());
    outputSchema = Schema.recordOf("user.flagged", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("dupe", Schema.of(Schema.Type.BOOLEAN)));
    final Set<StructuredRecord> expectedJoined2 = ImmutableSet.of(StructuredRecord.builder(outputSchema).set("id", 1L).set("name", "Samuel").set("dupe", true).build(), StructuredRecord.builder(outputSchema).set("id", 2L).set("name", "Dwayne").set("dupe", true).build(), StructuredRecord.builder(outputSchema).set("id", 3L).set("name", "Terry").set("dupe", false).build());
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink1.flush();
            sink2.flush();
            Set<StructuredRecord> actualAggs = new HashSet<>();
            Set<StructuredRecord> actualJoined = new HashSet<>();
            actualAggs.addAll(MockSink.readOutput(sink1));
            actualJoined.addAll(MockSink.readOutput(sink2));
            return expectedAggregates2.equals(actualAggs) && expectedJoined2.equals(actualJoined);
        }
    }, 1, TimeUnit.MINUTES);
    sparkManager.stop();
}

Example 4 with DataStreamsConfig

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

the class DataStreamsTest method testJoin.

@Test
public void testJoin() throws Exception {
    /*
     * source1 ----> t1 ------
     *                        | --> innerjoin ----> t4 ------
     * source2 ----> t2 ------                                 |
     *                                                         | ---> outerjoin --> sink1
     *                                                         |
     * source3 -------------------- t3 ------------------------
     */
    Schema inputSchema1 = Schema.recordOf("customerRecord", Schema.Field.of("customer_id", Schema.of(Schema.Type.STRING)), Schema.Field.of("customer_name", Schema.of(Schema.Type.STRING)));
    Schema inputSchema2 = Schema.recordOf("itemRecord", Schema.Field.of("item_id", Schema.of(Schema.Type.STRING)), Schema.Field.of("item_price", Schema.of(Schema.Type.LONG)), Schema.Field.of("cust_id", Schema.of(Schema.Type.STRING)), Schema.Field.of("cust_name", Schema.of(Schema.Type.STRING)));
    Schema inputSchema3 = Schema.recordOf("transactionRecord", Schema.Field.of("t_id", Schema.of(Schema.Type.STRING)), Schema.Field.of("c_id", Schema.of(Schema.Type.STRING)), Schema.Field.of("i_id", Schema.of(Schema.Type.STRING)));
    Schema outSchema2 = Schema.recordOf("join.output", Schema.Field.of("t_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("c_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("i_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("customer_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("customer_name", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("item_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("item_price", Schema.nullableOf(Schema.of(Schema.Type.LONG))), Schema.Field.of("cust_id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("cust_name", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    StructuredRecord recordSamuel = StructuredRecord.builder(inputSchema1).set("customer_id", "1").set("customer_name", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(inputSchema1).set("customer_id", "2").set("customer_name", "bob").build();
    StructuredRecord recordJane = StructuredRecord.builder(inputSchema1).set("customer_id", "3").set("customer_name", "jane").build();
    StructuredRecord recordCar = StructuredRecord.builder(inputSchema2).set("item_id", "11").set("item_price", 10000L).set("cust_id", "1").set("cust_name", "samuel").build();
    StructuredRecord recordBike = StructuredRecord.builder(inputSchema2).set("item_id", "22").set("item_price", 100L).set("cust_id", "3").set("cust_name", "jane").build();
    StructuredRecord recordTrasCar = StructuredRecord.builder(inputSchema3).set("t_id", "1").set("c_id", "1").set("i_id", "11").build();
    StructuredRecord recordTrasBike = StructuredRecord.builder(inputSchema3).set("t_id", "2").set("c_id", "3").set("i_id", "22").build();
    StructuredRecord recordTrasPlane = StructuredRecord.builder(inputSchema3).set("t_id", "3").set("c_id", "4").set("i_id", "33").build();
    List<StructuredRecord> input1 = ImmutableList.of(recordSamuel, recordBob, recordJane);
    List<StructuredRecord> input2 = ImmutableList.of(recordCar, recordBike);
    List<StructuredRecord> input3 = ImmutableList.of(recordTrasCar, recordTrasBike, recordTrasPlane);
    String outputName = "multiJoinOutputSink";
    DataStreamsConfig etlConfig = DataStreamsConfig.builder().addStage(new ETLStage("source1", MockSource.getPlugin(inputSchema1, input1))).addStage(new ETLStage("source2", MockSource.getPlugin(inputSchema2, input2))).addStage(new ETLStage("source3", MockSource.getPlugin(inputSchema3, input3))).addStage(new ETLStage("t1", IdentityTransform.getPlugin())).addStage(new ETLStage("t2", IdentityTransform.getPlugin())).addStage(new ETLStage("t3", IdentityTransform.getPlugin())).addStage(new ETLStage("t4", IdentityTransform.getPlugin())).addStage(new ETLStage("innerjoin", MockJoiner.getPlugin("t1.customer_id=t2.cust_id", "t1,t2", ""))).addStage(new ETLStage("outerjoin", MockJoiner.getPlugin("t4.item_id=t3.i_id", "", ""))).addStage(new ETLStage("multijoinSink", MockSink.getPlugin(outputName))).addConnection("source1", "t1").addConnection("source2", "t2").addConnection("source3", "t3").addConnection("t1", "innerjoin").addConnection("t2", "innerjoin").addConnection("innerjoin", "t4").addConnection("t3", "outerjoin").addConnection("t4", "outerjoin").addConnection("outerjoin", "multijoinSink").setBatchInterval("5s").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("JoinerApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    StructuredRecord joinRecordSamuel = StructuredRecord.builder(outSchema2).set("customer_id", "1").set("customer_name", "samuel").set("item_id", "11").set("item_price", 10000L).set("cust_id", "1").set("cust_name", "samuel").set("t_id", "1").set("c_id", "1").set("i_id", "11").build();
    StructuredRecord joinRecordJane = StructuredRecord.builder(outSchema2).set("customer_id", "3").set("customer_name", "jane").set("item_id", "22").set("item_price", 100L).set("cust_id", "3").set("cust_name", "jane").set("t_id", "2").set("c_id", "3").set("i_id", "22").build();
    StructuredRecord joinRecordPlane = StructuredRecord.builder(outSchema2).set("t_id", "3").set("c_id", "4").set("i_id", "33").build();
    final Set<StructuredRecord> expected = ImmutableSet.of(joinRecordSamuel, joinRecordJane, joinRecordPlane);
    final DataSetManager<Table> outputManager = getDataset(outputName);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            outputManager.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(outputManager));
            return expected.equals(outputRecords);
        }
    }, 4, TimeUnit.MINUTES);
    sparkManager.stop();
    sparkManager.waitForStatus(false, 10, 1);
    validateMetric(appId, "source1.records.out", 3);
    validateMetric(appId, "source2.records.out", 2);
    validateMetric(appId, "source3.records.out", 3);
    validateMetric(appId, "t1.records.in", 3);
    validateMetric(appId, "t1.records.out", 3);
    validateMetric(appId, "t2.records.in", 2);
    validateMetric(appId, "t2.records.out", 2);
    validateMetric(appId, "t3.records.in", 3);
    validateMetric(appId, "t3.records.out", 3);
    validateMetric(appId, "t4.records.in", 2);
    validateMetric(appId, "t4.records.out", 2);
    validateMetric(appId, "innerjoin.records.in", 5);
    validateMetric(appId, "innerjoin.records.out", 2);
    validateMetric(appId, "outerjoin.records.in", 5);
    validateMetric(appId, "outerjoin.records.out", 3);
    validateMetric(appId, "multijoinSink.records.in", 3);
}

Example 5 with DataStreamsConfig

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

the class DataStreamsTest method testSplitterTransform.

@Test
public void testSplitterTransform() throws Exception {
    Schema schema = Schema.recordOf("user", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("email", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    StructuredRecord user0 = StructuredRecord.builder(schema).set("id", 0L).build();
    StructuredRecord user1 = StructuredRecord.builder(schema).set("id", 1L).set("email", "one@example.com").build();
    StructuredRecord user2 = StructuredRecord.builder(schema).set("id", 2L).set("name", "two").build();
    StructuredRecord user3 = StructuredRecord.builder(schema).set("id", 3L).set("name", "three").set("email", "three@example.com").build();
    String sink1Name = "splitSink1";
    String sink2Name = "splitSink2";
    /*
     *
     *                                            |null --> sink1
     *                       |null--> splitter2 --|
     * source --> splitter1--|                    |non-null --|
     *                       |                                |--> sink2
     *                       |non-null------------------------|
     */
    DataStreamsConfig config = DataStreamsConfig.builder().setBatchInterval("5s").addStage(new ETLStage("source", MockSource.getPlugin(schema, ImmutableList.of(user0, user1, user2, user3)))).addStage(new ETLStage("splitter1", NullFieldSplitterTransform.getPlugin("name"))).addStage(new ETLStage("splitter2", NullFieldSplitterTransform.getPlugin("email"))).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1Name))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2Name))).addConnection("source", "splitter1").addConnection("splitter1", "splitter2", "null").addConnection("splitter1", "sink2", "non-null").addConnection("splitter2", "sink1", "null").addConnection("splitter2", "sink2", "non-null").build();
    AppRequest<DataStreamsConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("SplitterTest");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // run pipeline
    SparkManager sparkManager = appManager.getSparkManager(DataStreamsSparkLauncher.NAME);
    sparkManager.start();
    sparkManager.waitForStatus(true, 10, 1);
    // check output
    // sink1 should only have records where both name and email are null (user0)
    final DataSetManager<Table> sink1Manager = getDataset(sink1Name);
    final Set<StructuredRecord> expected1 = ImmutableSet.of(user0);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink1Manager.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sink1Manager));
            return expected1.equals(outputRecords);
        }
    }, 4, TimeUnit.MINUTES);
    // sink2 should have anything with a non-null name or non-null email
    final DataSetManager<Table> sink2Manager = getDataset(sink2Name);
    final Set<StructuredRecord> expected2 = ImmutableSet.of(user1, user2, user3);
    Tasks.waitFor(true, new Callable<Boolean>() {

        @Override
        public Boolean call() throws Exception {
            sink2Manager.flush();
            Set<StructuredRecord> outputRecords = new HashSet<>();
            outputRecords.addAll(MockSink.readOutput(sink2Manager));
            return expected2.equals(outputRecords);
        }
    }, 4, TimeUnit.MINUTES);
    sparkManager.stop();
    sparkManager.waitForStatus(false, 10, 1);
    validateMetric(appId, "source.records.out", 4);
    validateMetric(appId, "splitter1.records.in", 4);
    validateMetric(appId, "splitter1.records.out.non-null", 2);
    validateMetric(appId, "splitter1.records.out.null", 2);
    validateMetric(appId, "splitter2.records.in", 2);
    validateMetric(appId, "splitter2.records.out.non-null", 1);
    validateMetric(appId, "splitter2.records.out.null", 1);
    validateMetric(appId, "sink1.records.in", 1);
    validateMetric(appId, "sink2.records.in", 3);
}