Search in sources :

Example 1 with UpdateOptions

use of org.apache.samza.operators.UpdateOptions in project samza by apache.

the class SendToTableWithUpdateOperatorImpl method handleMessageAsync.

@Override
protected CompletionStage<Collection<KV<K, UpdateMessage<U, V>>>> handleMessageAsync(KV<K, UpdateMessage<U, V>> message, MessageCollector collector, TaskCoordinator coordinator) {
    final UpdateOptions options = spec.getUpdateOptions();
    final CompletableFuture<Void> updateFuture = table.updateAsync(message.getKey(), message.getValue().getUpdate());
    return updateFuture.handle((result, ex) -> {
        if (ex == null) {
            // success, no need to Put a default value
            return false;
        } else if (ex.getCause() instanceof RecordNotFoundException && message.getValue().hasDefault()) {
            // to PUT a default record for the key and then apply the update.
            if (options == UpdateOptions.UPDATE_WITH_DEFAULTS) {
                return true;
            } else {
                throw new SamzaException("Put default failed for update as the Update options was set to " + options + ". Please use UpdateOptions.UPDATE_WITH_DEFAULTS instead.");
            }
        } else {
            throw new SamzaException("Update failed with exception: ", ex);
        }
    }).thenCompose(shouldPutDefault -> {
        if (shouldPutDefault) {
            final CompletableFuture<Void> putFuture = table.putAsync(message.getKey(), message.getValue().getDefault());
            return putFuture.exceptionally(ex -> {
                LOG.warn("PUT default failed due to an exception. Ignoring the exception and proceeding with update. " + "The exception encountered is: ", ex);
                return null;
            }).thenCompose(res -> table.updateAsync(message.getKey(), message.getValue().getUpdate())).exceptionally(ex -> {
                throw new SamzaException("Update after Put default failed with exception: ", ex);
            });
        } else {
            return CompletableFuture.completedFuture(null);
        }
    }).thenApply(result -> Collections.singleton(message));
}
Also used : Logger(org.slf4j.Logger) UpdateMessage(org.apache.samza.operators.UpdateMessage) SendToTableWithUpdateOperatorSpec(org.apache.samza.operators.spec.SendToTableWithUpdateOperatorSpec) Collection(java.util.Collection) LoggerFactory(org.slf4j.LoggerFactory) CompletableFuture(java.util.concurrent.CompletableFuture) SamzaException(org.apache.samza.SamzaException) TaskCoordinator(org.apache.samza.task.TaskCoordinator) Context(org.apache.samza.context.Context) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) RemoteTable(org.apache.samza.table.remote.RemoteTable) CompletionStage(java.util.concurrent.CompletionStage) OperatorSpec(org.apache.samza.operators.spec.OperatorSpec) ReadWriteUpdateTable(org.apache.samza.table.ReadWriteUpdateTable) MessageCollector(org.apache.samza.task.MessageCollector) UpdateOptions(org.apache.samza.operators.UpdateOptions) CompactBatchProvider(org.apache.samza.table.batching.CompactBatchProvider) Collections(java.util.Collections) KV(org.apache.samza.operators.KV) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) CompletableFuture(java.util.concurrent.CompletableFuture) SamzaException(org.apache.samza.SamzaException) UpdateOptions(org.apache.samza.operators.UpdateOptions)

Example 2 with UpdateOptions

use of org.apache.samza.operators.UpdateOptions in project samza by apache.

the class TestRemoteTableEndToEnd method testSendToUpdatesWithoutUpdateOptions.

// Test will fail as we use sendTo with KV<K, UpdateMessage> stream without UpdateOptions
@Test(expected = SamzaException.class)
public void testSendToUpdatesWithoutUpdateOptions() throws Exception {
    // 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 InMemoryEnrichedPageViewWriteFunction2("testUpdateWithoutUpdateOptions", false)).withWriteRateLimit(1000);
        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<>());
        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);
    };
    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));
}
Also used : GuavaCacheTableDescriptor(org.apache.samza.table.descriptors.GuavaCacheTableDescriptor) Arrays(java.util.Arrays) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) ObjectInputStream(java.io.ObjectInputStream) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) RemoteTable(org.apache.samza.table.remote.RemoteTable) CachingTableDescriptor(org.apache.samza.table.descriptors.CachingTableDescriptor) Counter(org.apache.samza.metrics.Counter) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) NoOpTableReadFunction(org.apache.samza.table.remote.NoOpTableReadFunction) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Table(org.apache.samza.table.Table) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) MetricsRegistry(org.apache.samza.metrics.MetricsRegistry) Collectors(java.util.stream.Collectors) Executors(java.util.concurrent.Executors) Context(org.apache.samza.context.Context) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) TestRunner(org.apache.samza.test.framework.TestRunner) Matchers.any(org.mockito.Matchers.any) List(java.util.List) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) TestTableData.generateProfiles(org.apache.samza.test.table.TestTableData.generateProfiles) CacheBuilder(com.google.common.cache.CacheBuilder) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) UpdateMessage(org.apache.samza.operators.UpdateMessage) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) Profile(org.apache.samza.test.table.TestTableData.Profile) Function(java.util.function.Function) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) Matchers.anyString(org.mockito.Matchers.anyString) ArrayList(java.util.ArrayList) Base64Serializer(org.apache.samza.test.util.Base64Serializer) MockContext(org.apache.samza.context.MockContext) TableWriteFunction(org.apache.samza.table.remote.TableWriteFunction) ExpectedException(org.junit.rules.ExpectedException) Timer(org.apache.samza.metrics.Timer) BaseTableFunction(org.apache.samza.table.remote.BaseTableFunction) IOException(java.io.IOException) Test(org.junit.Test) SamzaException(org.apache.samza.SamzaException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Rule(org.junit.Rule) RateLimiter(org.apache.samza.util.RateLimiter) UpdateOptions(org.apache.samza.operators.UpdateOptions) Assert(org.junit.Assert) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) Matchers.anyString(org.mockito.Matchers.anyString) Profile(org.apache.samza.test.table.TestTableData.Profile) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) PageView(org.apache.samza.test.table.TestTableData.PageView) RemoteTable(org.apache.samza.table.remote.RemoteTable) Table(org.apache.samza.table.Table) StreamApplication(org.apache.samza.application.StreamApplication) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) KV(org.apache.samza.operators.KV) RateLimiter(org.apache.samza.util.RateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) NoOpSerde(org.apache.samza.serializers.NoOpSerde) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) Test(org.junit.Test)

Example 3 with UpdateOptions

use of org.apache.samza.operators.UpdateOptions in project samza by apache.

the class TestRemoteTableEndToEnd method testSendToWithDefaultsAndUpdateOnly.

// Test fails with the following exception:
// org.apache.samza.SamzaException: Put default failed for update as the UpdateOptions was set to UPDATE_ONLY.
// Please use UpdateOptions.UPDATE_WITH_DEFAULTS instead.
@Test(expected = SamzaException.class)
public void testSendToWithDefaultsAndUpdateOnly() throws Exception {
    String testName = "testSendToWithDefaultsAndUpdateOnly";
    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, false)).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<>());
        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_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));
}
Also used : GuavaCacheTableDescriptor(org.apache.samza.table.descriptors.GuavaCacheTableDescriptor) Arrays(java.util.Arrays) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) ObjectInputStream(java.io.ObjectInputStream) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) RemoteTable(org.apache.samza.table.remote.RemoteTable) CachingTableDescriptor(org.apache.samza.table.descriptors.CachingTableDescriptor) Counter(org.apache.samza.metrics.Counter) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) NoOpTableReadFunction(org.apache.samza.table.remote.NoOpTableReadFunction) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Table(org.apache.samza.table.Table) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) MetricsRegistry(org.apache.samza.metrics.MetricsRegistry) Collectors(java.util.stream.Collectors) Executors(java.util.concurrent.Executors) Context(org.apache.samza.context.Context) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) TestRunner(org.apache.samza.test.framework.TestRunner) Matchers.any(org.mockito.Matchers.any) List(java.util.List) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) TestTableData.generateProfiles(org.apache.samza.test.table.TestTableData.generateProfiles) CacheBuilder(com.google.common.cache.CacheBuilder) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) UpdateMessage(org.apache.samza.operators.UpdateMessage) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) Profile(org.apache.samza.test.table.TestTableData.Profile) Function(java.util.function.Function) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) Matchers.anyString(org.mockito.Matchers.anyString) ArrayList(java.util.ArrayList) Base64Serializer(org.apache.samza.test.util.Base64Serializer) MockContext(org.apache.samza.context.MockContext) TableWriteFunction(org.apache.samza.table.remote.TableWriteFunction) ExpectedException(org.junit.rules.ExpectedException) Timer(org.apache.samza.metrics.Timer) BaseTableFunction(org.apache.samza.table.remote.BaseTableFunction) IOException(java.io.IOException) Test(org.junit.Test) SamzaException(org.apache.samza.SamzaException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Rule(org.junit.Rule) RateLimiter(org.apache.samza.util.RateLimiter) UpdateOptions(org.apache.samza.operators.UpdateOptions) Assert(org.junit.Assert) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) Matchers.anyString(org.mockito.Matchers.anyString) Profile(org.apache.samza.test.table.TestTableData.Profile) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) List(java.util.List) ArrayList(java.util.ArrayList) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) PageView(org.apache.samza.test.table.TestTableData.PageView) RemoteTable(org.apache.samza.table.remote.RemoteTable) Table(org.apache.samza.table.Table) StreamApplication(org.apache.samza.application.StreamApplication) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) KV(org.apache.samza.operators.KV) RateLimiter(org.apache.samza.util.RateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) NoOpSerde(org.apache.samza.serializers.NoOpSerde) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) Test(org.junit.Test)

Aggregations

CompletableFuture (java.util.concurrent.CompletableFuture)3 SamzaException (org.apache.samza.SamzaException)3 Context (org.apache.samza.context.Context)3 KV (org.apache.samza.operators.KV)3 UpdateMessage (org.apache.samza.operators.UpdateMessage)3 UpdateOptions (org.apache.samza.operators.UpdateOptions)3 RecordNotFoundException (org.apache.samza.table.RecordNotFoundException)3 RemoteTable (org.apache.samza.table.remote.RemoteTable)3 CacheBuilder (com.google.common.cache.CacheBuilder)2 IOException (java.io.IOException)2 ObjectInputStream (java.io.ObjectInputStream)2 Duration (java.time.Duration)2 ArrayList (java.util.ArrayList)2 Arrays (java.util.Arrays)2 HashMap (java.util.HashMap)2 List (java.util.List)2 Map (java.util.Map)2 Executors (java.util.concurrent.Executors)2 TimeUnit (java.util.concurrent.TimeUnit)2 AtomicInteger (java.util.concurrent.atomic.AtomicInteger)2