Examples with GenericInputDescriptor org.apache.samza.system.descriptors.GenericInputDescriptor used on opensource projects

Example 21 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableWithBatchEndToEnd method doTestStreamTableJoinRemoteTablePartialUpdates.

private void doTestStreamTableJoinRemoteTablePartialUpdates(String testName, boolean isCompactBatch) throws Exception {
    final InMemoryWriteFunction writer = new InMemoryWriteFunction(testName);
    BATCH_READS.put(testName, new AtomicInteger());
    BATCH_WRITES.put(testName, new AtomicInteger());
    WRITTEN_RECORDS.put(testName, new HashMap<>());
    int count = 16;
    int batchSize = 4;
    String profiles = Base64Serializer.serialize(generateProfiles(count));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final RateLimiter writeRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        RemoteTableDescriptor<Integer, Profile, Void> inputTableDesc = new RemoteTableDescriptor<>("profile-table-1");
        inputTableDesc.withReadFunction(InMemoryReadFunction.getInMemoryReadFunction(testName, profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        // dummy reader
        TableReadFunction<Integer, EnrichedPageView> readFn = new MyReadFunction();
        RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView> outputTableDesc = new RemoteTableDescriptor<>("enriched-page-view-table-1");
        outputTableDesc.withReadFunction(readFn).withWriteFunction(writer).withRateLimiter(writeRateLimiter, creditFunction, creditFunction);
        if (isCompactBatch) {
            outputTableDesc.withBatchProvider(new CompactBatchProvider<Integer, EnrichedPageView, EnrichedPageView>().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        } else {
            outputTableDesc.withBatchProvider(new CompleteBatchProvider<Integer, EnrichedPageView, EnrichedPageView>().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        }
        Table<KV<Integer, EnrichedPageView>> table = appDesc.getTable(outputTableDesc);
        Table<KV<Integer, Profile>> inputTable = appDesc.getTable(inputTableDesc);
        DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(inputTable, new PageViewToProfileJoinFunction()).map(m -> new KV<>(m.getMemberId(), UpdateMessage.of(m, m))).sendTo(table, UpdateOptions.UPDATE_WITH_DEFAULTS);
    };
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    TestRunner.of(app).addInputStream(inputDescriptor, Arrays.asList(generatePageViewsWithDistinctKeys(count))).addConfig("task.max.concurrency", String.valueOf(count)).addConfig("task.async.commit", String.valueOf(true)).run(Duration.ofSeconds(10));
    Assert.assertEquals(count, WRITTEN_RECORDS.get(testName).size());
    Assert.assertNotNull(WRITTEN_RECORDS.get(testName).get(0));
    Assert.assertEquals(count / batchSize, BATCH_WRITES.get(testName).get());
}

Example 22 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method doTestStreamTableJoinRemoteTable.

private void doTestStreamTableJoinRemoteTable(boolean withCache, boolean defaultCache, boolean withUpdate, String testName) throws Exception {
    WRITTEN_RECORDS.put(testName, new ArrayList<>());
    // max member id for page views is 10
    final String profiles = Base64Serializer.serialize(generateProfiles(10));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        final RemoteTableDescriptor joinTableDesc = new RemoteTableDescriptor<Integer, TestTableData.Profile, Void>("profile-table-1").withReadFunction(InMemoryProfileReadFunction.getInMemoryReadFunction(profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        final RemoteTableDescriptor outputTableDesc = new RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView>("enriched-page-view-table-1").withReadFunction(new NoOpTableReadFunction<>()).withReadRateLimiterDisabled().withWriteFunction(new InMemoryEnrichedPageViewWriteFunction(testName)).withWriteRateLimit(1000);
        final Table<KV<Integer, Profile>> outputTable = withCache ? getCachingTable(outputTableDesc, defaultCache, appDesc) : appDesc.getTable(outputTableDesc);
        final Table<KV<Integer, Profile>> joinTable = withCache ? getCachingTable(joinTableDesc, defaultCache, appDesc) : appDesc.getTable(joinTableDesc);
        final DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        final GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        if (withUpdate) {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m, m))).sendTo(outputTable, UpdateOptions.UPDATE_WITH_DEFAULTS);
        } else {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> KV.of(m.getMemberId(), m)).sendTo(outputTable);
        }
    };
    int numPageViews = 40;
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    TestRunner.of(app).addInputStream(inputDescriptor, TestTableData.generatePartitionedPageViews(numPageViews, 4)).run(Duration.ofSeconds(10));
    Assert.assertEquals(numPageViews, WRITTEN_RECORDS.get(testName).size());
    Assert.assertNotNull(WRITTEN_RECORDS.get(testName).get(0));
    WRITTEN_RECORDS.get(testName).forEach(epv -> Assert.assertFalse(epv.company.contains("-")));
}

Example 23 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method doTestStreamTableJoinRemoteTableWithFirstTimeUpdates.

private void doTestStreamTableJoinRemoteTableWithFirstTimeUpdates(String testName, boolean withDefaults, boolean failUpdatesAlways) throws IOException {
    final String profiles = Base64Serializer.serialize(generateProfiles(30));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        final RemoteTableDescriptor joinTableDesc = new RemoteTableDescriptor<Integer, TestTableData.Profile, Void>("profile-table-1").withReadFunction(InMemoryProfileReadFunction.getInMemoryReadFunction(profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        final RemoteTableDescriptor outputTableDesc = new RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView>("enriched-page-view-table-1").withReadFunction(new NoOpTableReadFunction<>()).withReadRateLimiterDisabled().withWriteFunction(new InMemoryEnrichedPageViewWriteFunction2(testName, failUpdatesAlways)).withWriteRateLimit(1000);
        // counters to count puts and updates
        COUNTERS.put(testName + "-put", new AtomicInteger());
        COUNTERS.put(testName + "-update", new AtomicInteger());
        final Table<KV<Integer, Profile>> outputTable = appDesc.getTable(outputTableDesc);
        final Table<KV<Integer, Profile>> joinTable = appDesc.getTable(joinTableDesc);
        final DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        final GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        if (withDefaults) {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m, m))).sendTo(outputTable, UpdateOptions.UPDATE_WITH_DEFAULTS);
        } else {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m))).sendTo(outputTable, UpdateOptions.UPDATE_ONLY);
        }
    };
    int numPageViews = 15;
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    Map<Integer, List<PageView>> integerListMap = TestTableData.generatePartitionedPageViews(numPageViews, 1);
    TestRunner.of(app).addInputStream(inputDescriptor, integerListMap).run(Duration.ofSeconds(10));
    if (withDefaults) {
        Assert.assertEquals(10, COUNTERS.get(testName + "-put").intValue());
        Assert.assertEquals(15, COUNTERS.get(testName + "-update").intValue());
    }
}

Example 24 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestLocalTableEndToEnd method testSendTo.

@Test
public void testSendTo() {
    MyMapFunction mapFn = new MyMapFunction();
    StreamApplication app = appDesc -> {
        Table<KV<Integer, Profile>> table = appDesc.getTable(new InMemoryTableDescriptor<>("t1", KVSerde.of(new IntegerSerde(), new ProfileJsonSerde())));
        DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor(SYSTEM_NAME);
        GenericInputDescriptor<Profile> isd = ksd.getInputDescriptor(PROFILE_STREAM, new NoOpSerde<>());
        appDesc.getInputStream(isd).map(mapFn).sendTo(table);
    };
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor(SYSTEM_NAME);
    InMemoryInputDescriptor<Profile> profileStreamDesc = isd.getInputDescriptor(PROFILE_STREAM, new NoOpSerde<>());
    int numProfilesPerPartition = 10;
    int numInputPartitions = 4;
    Map<Integer, List<Profile>> inputProfiles = TestTableData.generatePartitionedProfiles(numProfilesPerPartition * numInputPartitions, numInputPartitions);
    TestRunner.of(app).addInputStream(profileStreamDesc, inputProfiles).run(Duration.ofSeconds(10));
    for (int i = 0; i < numInputPartitions; i++) {
        MyMapFunction mapFnCopy = MyMapFunction.getMapFunctionByTask(String.format("Partition %d", i));
        assertEquals(numProfilesPerPartition, mapFnCopy.received.size());
        mapFnCopy.received.forEach(p -> assertNotNull(mapFnCopy.table.get(p.getMemberId())));
    }
}

Example 25 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class SystemConsumerWithSamzaBench method start.

public void start() throws IOException, InterruptedException {
    super.start();
    MessageConsumer consumeFn = new MessageConsumer();
    StreamApplication app = appDesc -> {
        String systemFactoryName = new SystemConfig(config).getSystemFactory(systemName).get();
        GenericSystemDescriptor sd = new GenericSystemDescriptor(systemName, systemFactoryName);
        GenericInputDescriptor<Object> isd = sd.getInputDescriptor(streamId, new NoOpSerde<>());
        MessageStream<Object> stream = appDesc.getInputStream(isd);
        stream.map(consumeFn);
    };
    ApplicationRunner runner = ApplicationRunners.getApplicationRunner(app, new MapConfig());
    runner.run();
    while (consumeFn.getEventsConsumed() < totalEvents) {
        Thread.sleep(10);
    }
    Instant endTime = Instant.now();
    runner.kill();
    System.out.println("\n*******************");
    System.out.println(String.format("Started at %s Ending at %s ", consumeFn.startTime, endTime));
    System.out.println(String.format("Event Rate is %s Messages/Sec ", consumeFn.getEventsConsumed() * 1000 / Duration.between(consumeFn.startTime, Instant.now()).toMillis()));
    System.out.println("Event Rate is " + consumeFn.getEventsConsumed() * 1000 / Duration.between(consumeFn.startTime, endTime).toMillis());
    System.out.println("*******************\n");
    System.exit(0);
}