public void ApplyInternalChangelogTopicsWithExistingTopics()
{
AdminClientConfig config = new AdminClientConfig();
config.BootstrapServers = "localhost:9092";
StreamConfig config2 = new StreamConfig();
DefaultTopicManager manager = new DefaultTopicManager(config2, kafkaSupplier.GetAdmin(config));
((SyncProducer)kafkaSupplier.GetProducer(new ProducerConfig())).CreateTopic("topic");
IDictionary <string, InternalTopicConfig> topics = new Dictionary <string, InternalTopicConfig>();
topics.Add("topic", new UnwindowedChangelogTopicConfig
{
Name = "topic",
NumberPartitions = 1
});
topics.Add("topic1", new UnwindowedChangelogTopicConfig
{
Name = "topic1",
NumberPartitions = 1
});
var r = manager.ApplyAsync(0, topics).GetAwaiter().GetResult().ToList();
Assert.AreEqual(1, r.Count);
Assert.AreEqual("topic1", r[0]);
}
public void Initialize()
{
streamMetricsRegistry
= new StreamMetricsRegistry(Guid.NewGuid().ToString(),
MetricsRecordingLevel.DEBUG);
config.ApplicationId = "test-stream-thread";
config.StateDir = Guid.NewGuid().ToString();
config.Guarantee = ProcessingGuarantee.AT_LEAST_ONCE;
config.PollMs = 10;
config.CommitIntervalMs = 1;
var builder = new StreamBuilder();
var stream = builder.Stream <string, string>("topic");
stream.GroupByKey().Count();
stream.To("topic2");
var topo = builder.Build();
id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = builder.Build().Builder.BuildTopology(id);
syncKafkaSupplier = new SyncKafkaSupplier();
var producer = syncKafkaSupplier.GetProducer(config.ToProducerConfig());
var consumer = syncKafkaSupplier.GetConsumer(config.ToConsumerConfig(), null);
topicPartition = new TopicPartition("topic", 0);
task = new StreamTask(
threadId,
id,
new List <TopicPartition> {
topicPartition
},
processorTopology,
consumer,
config,
syncKafkaSupplier,
null,
new MockChangelogRegister(),
streamMetricsRegistry);
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
task.RestorationIfNeeded();
var activeRestorationSensor = streamMetricsRegistry.GetSensors().FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.ACTIVE_RESTORATION)));
Assert.AreEqual(1,
activeRestorationSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ACTIVE_RESTORATION,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
task.CompleteRestoration();
Assert.AreEqual(0,
activeRestorationSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ACTIVE_RESTORATION,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
}
public void TaskManagerCommitWithoutCommitNeeed()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
var builder = new StreamBuilder();
builder.Stream <string, string>("topic").To("topic2");
var topology = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var taskCreator = new TaskCreator(topology.Builder, config, "thread-0", supplier, producer);
var taskManager = new TaskManager(topology.Builder, taskCreator, supplier.GetAdmin(config.ToAdminConfig("admin")), consumer);
taskManager.CreateTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 0),
new TopicPartition("topic", 1),
new TopicPartition("topic", 2),
new TopicPartition("topic", 3),
});
Assert.AreEqual(4, taskManager.ActiveTasks.Count());
Assert.AreEqual(0, taskManager.CommitAll());
taskManager.Close();
}
public void WithNullMaterialize()
{
// CERTIFIED THAT SAME IF Materialize is null, a state store exist for count processor with a generated namestore
var config = new StreamConfig <StringSerDes, StringSerDes>();
var serdes = new StringSerDes();
config.ApplicationId = "test-count";
var builder = new StreamBuilder();
Materialized <string, long, IKeyValueStore <Bytes, byte[]> > m = null;
builder
.Table <string, string>("topic")
.GroupBy((k, v) => KeyValuePair.Create(k.ToUpper(), v))
.Count(m);
var topology = builder.Build();
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
null);
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
Assert.AreEqual(2, task.Context.States.StateStoreNames.Count());
var nameStore1 = task.Context.States.StateStoreNames.ElementAt(0);
var nameStore2 = task.Context.States.StateStoreNames.ElementAt(1);
Assert.IsNotNull(nameStore1);
Assert.IsNotNull(nameStore2);
Assert.AreNotEqual(string.Empty, nameStore1);
Assert.AreNotEqual(string.Empty, nameStore2);
var store1 = task.GetStore(nameStore1);
var store2 = task.GetStore(nameStore2);
Assert.IsInstanceOf <TimestampedKeyValueStore <string, string> >(store1);
Assert.IsInstanceOf <TimestampedKeyValueStore <string, long> >(store2);
Assert.AreEqual(0, (store1 as TimestampedKeyValueStore <string, string>).ApproximateNumEntries());
Assert.AreEqual(0, (store2 as TimestampedKeyValueStore <string, long>).ApproximateNumEntries());
}
public void GetCurrentPartitionMetadataTests()
{
var source = new System.Threading.CancellationTokenSource();
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.Guarantee = ProcessingGuarantee.AT_LEAST_ONCE;
config.PollMs = 1;
config.FollowMetadata = true;
var configConsumer = config.Clone();
configConsumer.ApplicationId = "test-consumer";
int?h = null;
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder
.Stream <string, string>("topic")
.MapValues((v) =>
{
h = StreamizMetadata.GetCurrentPartitionMetadata();
return(v);
})
.To("output");
var topo = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(configConsumer.ToConsumerConfig(), null);
var thread = StreamThread.Create(
"thread-0", "c0",
topo.Builder, config,
supplier, supplier.GetAdmin(config.ToAdminConfig("admin")),
0) as StreamThread;
thread.Start(source.Token);
producer.Produce("topic", new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1", new SerializationContext()),
Value = serdes.Serialize("coucou", new SerializationContext())
});
consumer.Subscribe("output");
ConsumeResult <byte[], byte[]> result = null;
do
{
result = consumer.Consume(100);
} while (result == null);
source.Cancel();
thread.Dispose();
Assert.NotNull(h);
Assert.AreEqual(0, h);
}
public void GetElementInStateStore()
{
var timeout = TimeSpan.FromSeconds(10);
var source = new CancellationTokenSource();
bool isRunningState = false;
DateTime dt = DateTime.Now;
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.PollMs = 10;
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var builder = new StreamBuilder();
builder.Table("topic", InMemory <string, string> .As("store"));
var t = builder.Build();
var stream = new KafkaStream(t, config, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
stream.Start(source.Token);
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
Assert.IsTrue(isRunningState);
if (isRunningState)
{
var serdes = new StringSerDes();
producer.Produce("topic",
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1"),
Value = serdes.Serialize("coucou")
});
Thread.Sleep(50);
var store = stream.Store(StoreQueryParameters.FromNameAndType("store", QueryableStoreTypes.KeyValueStore <string, string>()));
Assert.IsNotNull(store);
Assert.AreEqual(1, store.ApproximateNumEntries());
var item = store.Get("key1");
Assert.IsNotNull(item);
Assert.AreEqual("coucou", item);
}
source.Cancel();
stream.Close();
}
public async Task BuildGlobalStateStore()
{
var timeout = TimeSpan.FromSeconds(10);
bool isRunningState = false;
DateTime dt = DateTime.Now;
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.BootstrapServers = "127.0.0.1";
config.PollMs = 1;
var builder = new StreamBuilder();
builder.GlobalTable <string, string>("test", InMemory <string, string> .As("store"));
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(new ProducerConfig());
var t = builder.Build();
var stream = new KafkaStream(t, config, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
await stream.StartAsync();
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
Assert.IsTrue(isRunningState);
if (isRunningState)
{
var stringSerdes = new StringSerDes();
producer.Produce("test",
new Message <byte[], byte[]>
{
Key = stringSerdes.Serialize("key", new SerializationContext()),
Value = stringSerdes.Serialize("value", new SerializationContext())
});
Thread.Sleep(250);
var store = stream.Store(StoreQueryParameters.FromNameAndType("store", QueryableStoreTypes.KeyValueStore <string, string>()));
Assert.IsNotNull(store);
Assert.AreEqual(1, store.ApproximateNumEntries());
}
stream.Dispose();
}
public void WithNullMaterialize()
{
// CERTIFIED THAT SAME IF Materialize is null, a state store exist for count processor with a generated namestore
var config = new StreamConfig <StringSerDes, StringSerDes>();
var serdes = new StringSerDes();
config.ApplicationId = "test-window-count";
var builder = new StreamBuilder();
Materialized <string, long, IWindowStore <Bytes, byte[]> > m = null;
builder
.Stream <string, string>("topic")
.GroupByKey()
.WindowedBy(TumblingWindowOptions.Of(2000))
.Count(m);
var topology = builder.Build();
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
null,
new MockChangelogRegister(),
new StreamMetricsRegistry());
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
task.RestorationIfNeeded();
task.CompleteRestoration();
Assert.AreEqual(1, task.Context.States.StateStoreNames.Count());
var nameStore = task.Context.States.StateStoreNames.ElementAt(0);
Assert.IsNotNull(nameStore);
Assert.AreNotEqual(string.Empty, nameStore);
var store = task.GetStore(nameStore);
Assert.IsInstanceOf <ITimestampedWindowStore <string, long> >(store);
Assert.AreEqual(0, (store as ITimestampedWindowStore <string, long>).All().ToList().Count);
}
public void TaskManagerAssignedUnknownPartitions()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Stream <string, string>("topic")
.Map((k, v) => KeyValuePair.Create(k.ToUpper(), v.ToUpper()))
.To("topic2");
var topology = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var restoreConsumer = supplier.GetRestoreConsumer(config.ToConsumerConfig());
var storeChangelogReader =
new StoreChangelogReader(config, restoreConsumer, "thread-0", new StreamMetricsRegistry());
var taskCreator = new TaskCreator(topology.Builder, config, "thread-0", supplier, producer,
storeChangelogReader, new StreamMetricsRegistry());
var taskManager = new TaskManager(topology.Builder, taskCreator,
supplier.GetAdmin(config.ToAdminConfig("admin")), consumer, storeChangelogReader);
taskManager.CreateTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 0),
new TopicPartition("topic", 1)
});
taskManager.RevokeTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 1)
});
taskManager.CreateTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 0),
new TopicPartition("topic", 1),
new TopicPartition("topic", 2)
});
taskManager.TryToCompleteRestoration();
Assert.AreEqual(3, taskManager.ActiveTasks.Count());
Assert.AreEqual(0, taskManager.RevokedTasks.Count());
taskManager.Close();
}
public void StreamThreadCommitIntervalWorkflow()
{
var source = new System.Threading.CancellationTokenSource();
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.Guarantee = ProcessingGuarantee.AT_LEAST_ONCE;
config.PollMs = 1;
config.CommitIntervalMs = 1;
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Stream <string, string>("topic").To("topic2");
var topo = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig("test-consum"), null);
consumer.Subscribe("topic2");
var thread = StreamThread.Create(
"thread-0", "c0",
topo.Builder, new StreamMetricsRegistry(), config,
supplier, supplier.GetAdmin(config.ToAdminConfig("admin")),
0) as StreamThread;
thread.Start(source.Token);
producer.Produce("topic", new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1", new SerializationContext()),
Value = serdes.Serialize("coucou", new SerializationContext())
});
//WAIT STREAMTHREAD PROCESS MESSAGE
System.Threading.Thread.Sleep(100);
var message = consumer.Consume(100);
Assert.AreEqual("key1", serdes.Deserialize(message.Message.Key, new SerializationContext()));
Assert.AreEqual("coucou", serdes.Deserialize(message.Message.Value, new SerializationContext()));
var offsets = thread.GetCommittedOffsets(new List <TopicPartition> {
new TopicPartition("topic", 0)
},
TimeSpan.FromSeconds(10)).ToList();
Assert.AreEqual(1, offsets.Count);
Assert.AreEqual(1, offsets[0].Offset.Value);
Assert.AreEqual(0, offsets[0].TopicPartition.Partition.Value);
Assert.AreEqual("topic", offsets[0].Topic);
source.Cancel();
thread.Dispose();
}
public void RestoreFromRecovery()
{
storeChangelogReader.Restore();
stateMgr.Checkpoint();
stateMgr.Close();
store.Init(context, store);
stateMgr.InitializeOffsetsFromCheckpoint();
var producer = supplier.GetProducer(config.ToProducerConfig());
producer.Produce(changelogTopic, CreateMessage(changelogTopic, "key3", "value3"));
producer.Produce(changelogTopic, CreateMessage(changelogTopic, "key4", "value4"));
restoreConsumer.Resume(new TopicPartition(changelogTopic, 0).ToSingle());
storeChangelogReader.Restore();
stateMgr.Checkpoint();
stateMgr.Close();
File.Delete(Path.Combine(config.StateDir, config.ApplicationId, "0-0", ".checkpoint"));
store.Init(context, store);
stateMgr.InitializeOffsetsFromCheckpoint();
restoreConsumer.Resume(new TopicPartition(changelogTopic, 0).ToSingle());
storeChangelogReader.Restore();
Assert.AreEqual(4, store.All().ToList().Count);
var metadata = storeChangelogReader.GetMetadata(new TopicPartition(changelogTopic, 0));
Assert.IsNotNull(metadata);
Assert.AreEqual(ChangelogState.COMPLETED, metadata.ChangelogState);
Assert.AreEqual(3, metadata.RestoreEndOffset);
Assert.AreEqual(2, metadata.TotalRestored);
}
public void Init()
{
stateDir = Path.Combine(".", Guid.NewGuid().ToString());
config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-storechangelog-app";
config.StateDir = stateDir;
config.PollMs = 100;
supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
restoreConsumer = supplier.GetRestoreConsumer(config.ToConsumerConfig());
var topicPart = new TopicPartition("topic", 0);
changelogTopic = "store-changelog-topic";
var changelogsTopics = new Dictionary <string, string>();
changelogsTopics.Add("store", changelogTopic);
var id = new TaskId
{
Id = 0,
Partition = 0
};
store = new RocksDbKeyValueStore("store");
storeChangelogReader =
new StoreChangelogReader(config, restoreConsumer, "thread-0", new StreamMetricsRegistry());
stateMgr = new ProcessorStateManager(
id,
topicPart.ToSingle(),
changelogsTopics,
storeChangelogReader,
new OffsetCheckpointFile(Path.Combine(config.StateDir, config.ApplicationId, $"{id.Id}-{id.Partition}"))
);
Mock <AbstractTask> moq = new Mock <AbstractTask>();
moq.Setup(t => t.Id).Returns(new TaskId {
Id = 0, Partition = 0
});
context = new ProcessorContext(moq.Object, config, stateMgr, new StreamMetricsRegistry());
store.Init(context, store);
producer.Produce(changelogTopic, CreateMessage(changelogTopic, "key1", "value1"));
producer.Produce(changelogTopic, CreateMessage(changelogTopic, "key2", "value2"));
}
public void WithNullMaterialize()
{
// CERTIFIED THAT SAME IF Materialize is null, a state store exist for count processor with a generated namestore
var config = new StreamConfig <StringSerDes, StringSerDes>();
var serdes = new StringSerDes();
config.ApplicationId = "test-window-count";
var builder = new StreamBuilder();
Materialized <string, int, IWindowStore <Bytes, byte[]> > m = null;
builder
.Stream <string, string>("topic")
.GroupByKey()
.WindowedBy(TumblingWindowOptions.Of(2000))
.Aggregate(
() => 0,
(k, v, agg) => Math.Max(v.Length, agg),
m);
var topology = builder.Build();
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
null);
task.GroupMetadata = consumer as SyncConsumer;
Assert.Throws <StreamsException>(() => task.InitializeStateStores());
}
public void RestoreFromCheckpoint()
{
var serdes = new StringSerDes();
byte[] GetBytes(string content)
=> serdes.Serialize(content, SerializationContext.Empty);
kvStoreMock
.Setup(s => s.Init(context, It.IsAny <IStateStore>()))
.Callback((ProcessorContext c, IStateStore store) => { stateManager.Register(store, (k, v, t) => { }); });
otherStoreMock
.Setup(s => s.Init(context, It.IsAny <IStateStore>()))
.Callback((ProcessorContext c, IStateStore store) => { stateManager.Register(store, (k, v, t) => { }); });
StringBuilder sb = new StringBuilder();
sb.AppendLine("0").AppendLine("1").AppendLine("kv-store-topic 0 10");
var producer = mockKafkaSupplier.GetProducer(new ProducerConfig());
for (int i = 0; i < 10; ++i)
{
producer.Produce("other-store-topic",
new Message <byte[], byte[]>()
{
Key = GetBytes($"key{i}"),
Value = GetBytes($"value{i}")
});
}
Directory.CreateDirectory(Path.Combine(streamConfigMock.Object.StateDir,
streamConfigMock.Object.ApplicationId, "global"));
File.WriteAllText(
Path.Combine(streamConfigMock.Object.StateDir, streamConfigMock.Object.ApplicationId, "global",
".checkpoint"),
sb.ToString());
stateManager.Initialize();
Assert.AreEqual(10, stateManager.ChangelogOffsets[new TopicPartition("kv-store-topic", 0)]);
Assert.AreEqual(9, stateManager.ChangelogOffsets[new TopicPartition("other-store-topic", 0)]);
Directory.Delete(Path.Combine(streamConfigMock.Object.StateDir), true);
}
public void Begin()
{
valueAndTimestampSerDes = new ValueAndTimestampSerDes <string>(stringSerDes);
config = new StreamConfig();
config.ApplicationId = $"unit-test-changelogging-tkv";
id = new TaskId {
Id = 0, Partition = 0
};
partition = new TopicPartition("source", 0);
kafkaSupplier = new SyncKafkaSupplier();
var producerConfig = new ProducerConfig();
producerConfig.ClientId = "producer-1";
var producerClient = kafkaSupplier.GetProducer(producerConfig);
recordCollector = new RecordCollector("p-1", config, id, new NoRunnableSensor("s", "s", MetricsRecordingLevel.DEBUG));
recordCollector.Init(ref producerClient);
var changelogsTopics = new Dictionary <string, string> {
{ "test-store", "test-store-changelog" }
};
stateManager = new ProcessorStateManager(
id,
new List <TopicPartition> {
partition
},
changelogsTopics,
new MockChangelogRegister(),
new MockOffsetCheckpointManager());
task = new Mock <AbstractTask>();
task.Setup(k => k.Id).Returns(id);
context = new ProcessorContext(task.Object, config, stateManager, new StreamMetricsRegistry());
context.UseRecordCollector(recordCollector);
var inmemorystore = new InMemoryKeyValueStore("test-store");
store = new ChangeLoggingTimestampedKeyValueBytesStore(inmemorystore);
store.Init(context, store);
}
public void WithNullMaterialize()
{
// CERTIFIED THAT SAME IF Materialize is null, a state store exist for count processor with a generated namestore
var config = new StreamConfig();
var serdes = new StringSerDes();
config.ApplicationId = "test-window-reduce";
var builder = new StreamBuilder();
builder
.Stream <string, string>("topic")
.GroupByKey()
.WindowedBy(TumblingWindowOptions.Of(2000))
.Reduce((v1, v2) => v1.Length > v2.Length ? v1 : v2);
var topology = builder.Build();
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
null,
new MockChangelogRegister(),
new StreamMetricsRegistry());
task.GroupMetadata = consumer as SyncConsumer;
Assert.Throws <StreamsException>(() => task.InitializeStateStores());
}
public static UnassignedStreamTask Create(TaskId id)
{
var config = new StreamConfig();
config.ApplicationId = "un-assigned-stream-task";
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
return(new UnassignedStreamTask(
"un-assigned-stream-task",
id,
new List <TopicPartition>(),
ProcessorTopology.EMPTY,
consumer,
config,
supplier,
producer));
}
public static UnassignedStreamTask Create()
{
var config = new StreamConfig();
config.ApplicationId = "un-assigned-stream-task";
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
return(new UnassignedStreamTask(
"un-assigned-stream-task",
new TaskId {
Id = -1, Partition = -1
},
new List <TopicPartition>(),
null,
consumer,
config,
supplier,
producer));
}
public async Task GetRangeKVStateStore()
{
var timeout = TimeSpan.FromSeconds(10);
bool isRunningState = false;
DateTime dt = DateTime.Now;
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.BootstrapServers = "127.0.0.1";
config.PollMs = 10;
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var builder = new StreamBuilder();
builder.Table("topic", InMemory <string, string> .As("store"));
var t = builder.Build();
var stream = new KafkaStream(t, config, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
await stream.StartAsync();
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
Assert.IsTrue(isRunningState);
if (isRunningState)
{
var serdes = new StringSerDes();
producer.Produce("topic",
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1", new SerializationContext()),
Value = serdes.Serialize("coucou", new SerializationContext())
});
Thread.Sleep(50);
var store = stream.Store(StoreQueryParameters.FromNameAndType("store", QueryableStoreTypes.KeyValueStore <string, string>()));
Assert.IsNotNull(store);
var list = store.Range("key1", "key2").ToList();
Assert.AreEqual(1, list.Count);
var item = list[0];
Assert.IsNotNull(item);
Assert.AreEqual("coucou", item.Value);
Assert.AreEqual("key1", item.Key);
}
stream.Dispose();
}
public void TaskManagerCommit()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Stream <string, string>("topic")
.Map((k, v) => KeyValuePair.Create(k.ToUpper(), v.ToUpper()))
.To("topic2");
var topology = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var taskCreator = new TaskCreator(topology.Builder, config, "thread-0", supplier, producer);
var taskManager = new TaskManager(topology.Builder, taskCreator, supplier.GetAdmin(config.ToAdminConfig("admin")), consumer);
taskManager.CreateTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 0),
new TopicPartition("topic", 1),
new TopicPartition("topic", 2),
new TopicPartition("topic", 3),
});
Assert.AreEqual(4, taskManager.ActiveTasks.Count());
var part = new TopicPartition("topic", 0);
var task = taskManager.ActiveTaskFor(part);
List <ConsumeResult <byte[], byte[]> > messages = new List <ConsumeResult <byte[], byte[]> >();
int offset = 0;
for (int i = 0; i < 5; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(part, offset++)
});
}
task.AddRecords(messages);
Assert.IsTrue(task.CanProcess(DateTime.Now.GetMilliseconds()));
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
}
// ONLY ONE TASK HAVE BEEN RECORDS
Assert.AreEqual(1, taskManager.CommitAll());
// CHECK IN TOPIC topic2
consumer.Subscribe("topic2");
List <ConsumeResult <byte[], byte[]> > results = new List <ConsumeResult <byte[], byte[]> >();
ConsumeResult <byte[], byte[]> result = null;
do
{
result = consumer.Consume(100);
if (result != null)
{
results.Add(result);
consumer.Commit(result);
}
} while (result != null);
Assert.AreEqual(5, results.Count);
for (int i = 0; i < 5; ++i)
{
Assert.AreEqual($"KEY{i + 1}", serdes.Deserialize(results[i].Message.Key, new SerializationContext()));
Assert.AreEqual($"VALUE{i+1}", serdes.Deserialize(results[i].Message.Value, new SerializationContext()));
}
// NO RECORD IN THIS TASKS
part = new TopicPartition("topic", 2);
task = taskManager.ActiveTaskFor(part);
Assert.IsFalse(task.CanProcess(DateTime.Now.GetMilliseconds()));
Assert.IsFalse(task.Process());
taskManager.Close();
}
public void StreamThreadNormalWorkflow()
{
bool metricsReporterCalled = false;
List <ThreadState> allStates = new List <ThreadState>();
var expectedStates = new List <ThreadState>
{
ThreadState.CREATED,
ThreadState.STARTING,
ThreadState.PARTITIONS_ASSIGNED,
ThreadState.RUNNING,
ThreadState.PENDING_SHUTDOWN,
ThreadState.DEAD
};
var source = new System.Threading.CancellationTokenSource();
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.Guarantee = ProcessingGuarantee.AT_LEAST_ONCE;
config.PollMs = 1;
config.MetricsReporter = (sensor) => { metricsReporterCalled = true; };
config.AddOrUpdate(StreamConfig.metricsIntervalMsCst, 10);
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Stream <string, string>("topic").To("topic2");
var topo = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig("test-consum"), null);
consumer.Subscribe("topic2");
var thread = StreamThread.Create(
"thread-0", "c0",
topo.Builder, new StreamMetricsRegistry(), config,
supplier, supplier.GetAdmin(config.ToAdminConfig("admin")),
0) as StreamThread;
allStates.Add(thread.State);
thread.StateChanged += (t, o, n) =>
{
Assert.IsInstanceOf <ThreadState>(n);
allStates.Add(n as ThreadState);
};
thread.Start(source.Token);
producer.Produce("topic", new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1", new SerializationContext()),
Value = serdes.Serialize("coucou", new SerializationContext())
});
//WAIT STREAMTHREAD PROCESS MESSAGE
System.Threading.Thread.Sleep(100);
var message = consumer.Consume(100);
source.Cancel();
thread.Dispose();
Assert.AreEqual("key1", serdes.Deserialize(message.Message.Key, new SerializationContext()));
Assert.AreEqual("coucou", serdes.Deserialize(message.Message.Value, new SerializationContext()));
Assert.AreEqual(expectedStates, allStates);
Assert.IsTrue(metricsReporterCalled);
}
//[Test]
// TODO : fix that
public void WorkflowCompleteBufferedRecordsTest()
{
int maxBuffered = 10;
var token = new System.Threading.CancellationTokenSource();
var serdes = new StringSerDes();
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-group";
config.MaxTaskIdleMs = (long)TimeSpan.FromSeconds(100).TotalMilliseconds;
config.BufferedRecordsPerPartition = maxBuffered;
config.PollMs = 10;
var builder = new StreamBuilder();
var stream1 = builder.Stream <string, string>("topic1");
var stream2 = builder.Stream <string, string>("topic2");
stream1
.Join(stream2, (v1, v2) => $"{v1}-{v2}", JoinWindowOptions.Of(TimeSpan.FromSeconds(10)))
.To("output");
var topo = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig("test-consum"), null);
consumer.Subscribe("output");
var thread = StreamThread.Create(
"thread-0", "c0",
topo.Builder, config,
supplier, supplier.GetAdmin(config.ToAdminConfig("admin")),
0) as StreamThread;
thread.Start(token.Token);
for (int i = 0; i < maxBuffered + 1; ++i)
{
producer.Produce("topic1", new Message <byte[], byte[]>
{
Key = serdes.Serialize("key", new SerializationContext()),
Value = serdes.Serialize($"coucou{i}", new SerializationContext())
});
}
// CONSUME PAUSE AFTER maxBuffered + 1 messages
System.Threading.Thread.Sleep(50);
// Add one message more with consumer in stream thread in pause
producer.Produce("topic1", new Message <byte[], byte[]>
{
Key = serdes.Serialize("key", new SerializationContext()),
Value = serdes.Serialize($"coucou{maxBuffered+1}", new SerializationContext())
});
Assert.AreEqual(1, thread.ActiveTasks.Count());
var task = thread.ActiveTasks.ToArray()[0];
Assert.IsNotNull(task.Grouper);
Assert.IsFalse(task.Grouper.AllPartitionsBuffered);
Assert.AreEqual(maxBuffered + 1, task.Grouper.NumBuffered());
Assert.AreEqual(maxBuffered + 1, task.Grouper.NumBuffered(new TopicPartition("topic1", 0)));
Assert.AreEqual(0, task.Grouper.NumBuffered(new TopicPartition("topic2", 0)));
producer.Produce("topic2", new Message <byte[], byte[]>
{
Key = serdes.Serialize("key", new SerializationContext()),
Value = serdes.Serialize($"test", new SerializationContext())
});
List <ConsumeResult <byte[], byte[]> > records = new List <ConsumeResult <byte[], byte[]> >();
do
{
records.AddRange(consumer.ConsumeRecords(TimeSpan.FromMilliseconds(100)).ToList());
} while (records.Count() <= 12);
Assert.AreEqual(maxBuffered + 2, records.Count());
for (int i = 0; i < maxBuffered + 2; ++i)
{
var message = records.ToArray()[i];
Assert.AreEqual("key", serdes.Deserialize(message.Message.Key, new SerializationContext()));
Assert.IsTrue(serdes.Deserialize(message.Message.Value, new SerializationContext()).Contains($"coucou{i}-"));
}
token.Cancel();
thread.Dispose();
}
public void TaskMetricsTest()
{
var serdes = new StringSerDes();
var cloneConfig = config.Clone();
cloneConfig.ApplicationId = "consume-test";
var producer = syncKafkaSupplier.GetProducer(cloneConfig.ToProducerConfig());
var consumer = syncKafkaSupplier.GetConsumer(cloneConfig.ToConsumerConfig("test-consum"), null);
consumer.Subscribe("topic2");
int nbMessage = 1000;
// produce 1000 messages to input topic
List <ConsumeResult <byte[], byte[]> > messages = new List <ConsumeResult <byte[], byte[]> >();
int offset = 0;
for (int i = 0; i < nbMessage; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(topicPartition, offset++)
});
}
task.AddRecords(messages);
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
Assert.IsTrue(task.CommitNeeded);
task.Commit();
}
var messagesSink = new List <ConsumeResult <byte[], byte[]> >();
AssertExtensions.WaitUntil(() =>
{
messagesSink.AddRange(consumer.ConsumeRecords(TimeSpan.FromSeconds(1)));
return(messagesSink.Count < nbMessage);
}, TimeSpan.FromSeconds(5),
TimeSpan.FromMilliseconds(10));
long now = DateTime.Now.GetMilliseconds();
var sensors = streamMetricsRegistry.GetThreadScopeSensor(threadId);
var processorSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.PROCESS)));
Assert.AreEqual(2, processorSensor.Metrics.Count());
Assert.AreEqual(nbMessage,
processorSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.PROCESS + StreamMetricsRegistry.TOTAL_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
Assert.IsTrue(
(double)(processorSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.PROCESS + StreamMetricsRegistry.RATE_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value) > 0d);
var enforcedProcessorSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.ENFORCED_PROCESSING)));
Assert.AreEqual(2, enforcedProcessorSensor.Metrics.Count());
Assert.AreEqual(0,
enforcedProcessorSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ENFORCED_PROCESSING + StreamMetricsRegistry.TOTAL_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
Assert.AreEqual(0,
enforcedProcessorSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ENFORCED_PROCESSING + StreamMetricsRegistry.RATE_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
var processLatency = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.PROCESS_LATENCY)));
Assert.AreEqual(2, processLatency.Metrics.Count());
Assert.IsTrue(
(double)processLatency.Metrics[MetricName.NameAndGroup(
TaskMetrics.PROCESS_LATENCY + StreamMetricsRegistry.AVG_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value > 0d);
Assert.IsTrue(
(double)processLatency.Metrics[MetricName.NameAndGroup(
TaskMetrics.PROCESS_LATENCY + StreamMetricsRegistry.MAX_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value > 0d);
var commitSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.COMMIT)));
Assert.AreEqual(2, commitSensor.Metrics.Count());
Assert.AreEqual(nbMessage,
commitSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.COMMIT + StreamMetricsRegistry.TOTAL_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
Assert.IsTrue(
(double)commitSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.COMMIT + StreamMetricsRegistry.RATE_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value > 0d);
var droppedRecordSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.DROPPED_RECORDS)));
Assert.AreEqual(2, droppedRecordSensor.Metrics.Count());
Assert.AreEqual(0,
droppedRecordSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.DROPPED_RECORDS + StreamMetricsRegistry.TOTAL_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
Assert.AreEqual(0,
droppedRecordSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.DROPPED_RECORDS + StreamMetricsRegistry.RATE_SUFFIX,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
var activeBufferedRecordSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.ACTIVE_TASK_PREFIX + TaskMetrics.BUFFER_COUNT)));
Assert.AreEqual(1, activeBufferedRecordSensor.Metrics.Count());
Assert.AreEqual(0,
activeBufferedRecordSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ACTIVE_TASK_PREFIX + TaskMetrics.BUFFER_COUNT,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
var restorationRecordsSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.RESTORATION_RECORDS)));
Assert.AreEqual(1, restorationRecordsSensor.Metrics.Count());
Assert.AreEqual(0,
restorationRecordsSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.RESTORATION_RECORDS,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
var activeRestorationSensor = sensors.FirstOrDefault(s => s.Name.Equals(GetSensorName(TaskMetrics.ACTIVE_RESTORATION)));
Assert.AreEqual(1, activeRestorationSensor.Metrics.Count());
Assert.AreEqual(0,
activeRestorationSensor.Metrics[MetricName.NameAndGroup(
TaskMetrics.ACTIVE_RESTORATION,
StreamMetricsRegistry.TASK_LEVEL_GROUP)].Value);
}
public void StreamTaskWrittingCheckpoint()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
config.StateDir = Path.Combine(".", Guid.NewGuid().ToString());
var serdes = new StringSerDes();
var builder = new StreamBuilder();
var table = builder.Table("topic", RocksDb <string, string> .As("store").WithLoggingEnabled());
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var topology = builder.Build();
topology.Builder.RewriteTopology(config);
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
producer,
new MockChangelogRegister()
, new StreamMetricsRegistry());
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
task.RestorationIfNeeded();
task.CompleteRestoration();
List <ConsumeResult <byte[], byte[]> > messages = new List <ConsumeResult <byte[], byte[]> >();
int offset = 0;
for (int i = 0; i < 5; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(part, offset++)
});
}
task.AddRecords(messages);
Assert.IsTrue(task.CanProcess(DateTime.Now.GetMilliseconds()));
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
Assert.IsTrue(task.CommitNeeded);
task.Commit();
}
task.MayWriteCheckpoint(true);
messages = new List <ConsumeResult <byte[], byte[]> >();
for (int i = 0; i < StateManagerTools.OFFSET_DELTA_THRESHOLD_FOR_CHECKPOINT + 10; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(part, offset++)
});
}
task.AddRecords(messages);
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
}
task.MayWriteCheckpoint(false);
var lines = File.ReadAllLines(Path.Combine(config.StateDir, config.ApplicationId, "0-0", ".checkpoint"));
Assert.AreEqual(3, lines.Length);
Assert.AreEqual("test-app-store-changelog 0 10014", lines[2]);
task.Suspend();
task.Close();
Directory.Delete(config.StateDir, true);
}
public void StreamTaskSuspendResume()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Table <string, string>("topic", InMemory <string, string> .As("store").WithLoggingDisabled())
.MapValues((k, v) => v.ToUpper())
.ToStream()
.To("topic2");
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var topology = builder.Build();
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
producer,
new MockChangelogRegister()
, new StreamMetricsRegistry());
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
task.RestorationIfNeeded();
task.CompleteRestoration();
List <ConsumeResult <byte[], byte[]> > messages = new List <ConsumeResult <byte[], byte[]> >();
int offset = 0;
for (int i = 0; i < 5; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(part, offset++)
});
}
task.AddRecords(messages);
Assert.IsTrue(task.CanProcess(DateTime.Now.GetMilliseconds()));
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
Assert.IsTrue(task.CommitNeeded);
task.Commit();
}
Assert.IsNotNull(task.GetStore("store"));
task.Suspend();
Assert.IsNull(task.GetStore("store"));
task.Resume();
task.RestorationIfNeeded();
Assert.IsNotNull(task.GetStore("store"));
task.AddRecords(messages);
Assert.IsTrue(task.CanProcess(DateTime.Now.GetMilliseconds()));
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
Assert.IsTrue(task.CommitNeeded);
task.Commit();
}
// CHECK IN TOPIC topic2
consumer.Subscribe("topic2");
List <ConsumeResult <byte[], byte[]> > results = new List <ConsumeResult <byte[], byte[]> >();
ConsumeResult <byte[], byte[]> result = null;
do
{
result = consumer.Consume(100);
if (result != null)
{
results.Add(result);
consumer.Commit(result);
}
} while (result != null);
Assert.AreEqual(10, results.Count);
task.Close();
}
public async Task GetWindowElementInStateStore()
{
var timeout = TimeSpan.FromSeconds(10);
bool isRunningState = false;
DateTime dt = DateTime.Now;
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test";
config.BootstrapServers = "127.0.0.1";
config.PollMs = 10;
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var builder = new StreamBuilder();
builder
.Stream <string, string>("test")
.GroupByKey()
.WindowedBy(TumblingWindowOptions.Of(TimeSpan.FromMinutes(1)))
.Count(InMemoryWindows <string, long> .As("store"));
var t = builder.Build();
var stream = new KafkaStream(t, config, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
await stream.StartAsync();
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
Assert.IsTrue(isRunningState);
if (isRunningState)
{
var serdes = new StringSerDes();
dt = DateTime.Now;
producer.Produce("test",
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("key1", new SerializationContext()),
Value = serdes.Serialize("coucou", new SerializationContext()),
Timestamp = new Confluent.Kafka.Timestamp(dt)
});
Thread.Sleep(50);
var store = stream.Store(StoreQueryParameters.FromNameAndType("store", QueryableStoreTypes.WindowStore <string, long>()));
Assert.IsNotNull(store);
var @enum = store.All();
Assert.AreEqual(1, store.All().ToList().Count);
var item = store.Fetch("key1", dt.AddMinutes(-1), dt.AddMinutes(1));
Assert.IsNotNull(item);
Assert.IsTrue(item.MoveNext());
Assert.IsTrue(item.Current.HasValue);
Assert.AreEqual(1, item.Current.Value.Value);
item.Dispose();
}
stream.Dispose();
}
public async Task exec(IConfiguration config, IServiceProvider services)
{
Console.WriteLine("Process");
var destTopic = config["spring.cloud.stream.bindings.output.destination"];
Console.WriteLine(destTopic);
using (var scope = services.CreateScope())
{
this._dataService = scope.ServiceProvider
.GetRequiredService <IDataService>();
bool isRunningState = false;
var timeout = TimeSpan.FromSeconds(10);
DateTime dt = DateTime.Now;
Order[] capture = this._dataService.readData();
// Inyectamos los datos obtenidos al Stream
var sConfig = new StreamConfig <StringSerDes, StringSerDes>();
sConfig.ApplicationId = config["SPRING_CLOUD_APPLICATION_GUID"];
sConfig.BootstrapServers = config["SPRING_CLOUD_STREAM_KAFKA_BINDER_BROKERS"];
sConfig.SchemaRegistryUrl = config["SchemaRegistryUrl"];
sConfig.AutoRegisterSchemas = true;
sConfig.NumStreamThreads = 10;
sConfig.Acks = Acks.All;
sConfig.AddConsumerConfig("allow.auto.create.topics", "true");
sConfig.InnerExceptionHandler = (e) => ExceptionHandlerResponse.CONTINUE;
var schemaRegistryClient = new CachedSchemaRegistryClient
(new SchemaRegistryConfig
{
Url = sConfig.SchemaRegistryUrl
});
var supplier = new SyncKafkaSupplier(new KafkaLoggerAdapter(sConfig));
var producerConfig = sConfig.ToProducerConfig();
var adminConfig = sConfig.ToAdminConfig(sConfig.ApplicationId);
var admin = supplier.GetAdmin(adminConfig);
// try
// {
// var topic = new TopicSpecification
// {
// Name = destTopic,
// NumPartitions = 1,
// ReplicationFactor = 3
// };
// var topicProduct = new TopicSpecification
// {
// Name = "product-external",
// NumPartitions = 1,
// ReplicationFactor = 3
// };
// IList<TopicSpecification> topics = new List<TopicSpecification>();
// topics.Add(topic);
// topics.Add(topicProduct);
// await admin.CreateTopicsAsync(topics);
// }
// catch (Exception topicExists)
// {
// Console.WriteLine("Topic alreade exists");
// Console.Write(topicExists);
// }
var producer = supplier.GetProducer(producerConfig);
StreamBuilder builder = new StreamBuilder();
var serdes = new SchemaAvroSerDes <Order>();
var keySerdes = new Int32SerDes();
builder.Table(destTopic, keySerdes, serdes, InMemory <int, Order> .As(config["table"]));
var t = builder.Build();
KafkaStream stream = new KafkaStream(t, sConfig, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
await stream.StartAsync();
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
if (isRunningState)
{
// //create a well formatted Endpoint in external topic
var endpProducer = new ProducerBuilder <byte[], Endpoint>(producerConfig)
.SetValueSerializer(new AvroSerializer <Endpoint>(schemaRegistryClient, new AvroSerializerConfig {
AutoRegisterSchemas = true
}).AsSyncOverAsync()).Build();
//create a well formatted Product in external topic
var productProducer = new ProducerBuilder <byte[], Product>(producerConfig)
.SetValueSerializer(new AvroSerializer <Product>(schemaRegistryClient, new AvroSerializerConfig {
AutoRegisterSchemas = true
}).AsSyncOverAsync()).Build();
for (int k = 1; k < 10; k++)
{
endpProducer.Produce("api-endpoints",
new Message <byte[], Endpoint>
{
Key = new Int32SerDes().Serialize(k, new SerializationContext()),
Value = new Endpoint
{
endpoint_id = ("endpoint" + k),
endpoint_url = ("http://endpoint" + k + "/"),
http_method = "POST"
}
}, (d) =>
{
if (d.Status == PersistenceStatus.Persisted)
{
Console.WriteLine("Endpoint Message sent !");
}
});
productProducer.Produce("product-external",
new Message <byte[], Product>
{
Key = new Int32SerDes().Serialize(1, new SerializationContext()),
Value = new Product
{
name = "Producto de Software",
price = 1234.5F,
product_id = 3
}
}, (d) =>
{
if (d.Status == PersistenceStatus.Persisted)
{
Console.WriteLine("Product Message sent !");
}
});
}
Thread.Sleep(10);
for (int k = 1; k < 10; k++)
{
producer.Produce(destTopic,
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = keySerdes.Serialize(k, new SerializationContext()),
Value = serdes.Serialize(new Order
{
order_id = k,
price = 123.5F,
product_id = k
}, new SerializationContext())
}, (d) =>
{
if (d.Status == PersistenceStatus.Persisted)
{
Console.WriteLine("Order Message sent !");
}
});
}
Thread.Sleep(50);
}
}
}
public void StandardWorkflowTaskManager()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
var builder = new StreamBuilder();
builder.Stream <string, string>("topic").To("topic2");
var topology = builder.Build();
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var taskCreator = new TaskCreator(topology.Builder, config, "thread-0", supplier, producer);
var taskManager = new TaskManager(topology.Builder, taskCreator, supplier.GetAdmin(config.ToAdminConfig("admin")), consumer);
taskManager.CreateTasks(
new List <TopicPartition>
{
new TopicPartition("topic", 0),
new TopicPartition("topic", 1),
new TopicPartition("topic", 2),
new TopicPartition("topic", 3),
});
Assert.AreEqual(4, taskManager.ActiveTasks.Count());
for (int i = 0; i < 4; ++i)
{
var task = taskManager.ActiveTaskFor(new TopicPartition("topic", i));
Assert.IsNotNull(task);
Assert.AreEqual("test-app", task.ApplicationId);
Assert.IsFalse(task.CanProcess(DateTime.Now.GetMilliseconds()));
Assert.IsFalse(task.CommitNeeded);
Assert.IsFalse(task.HasStateStores);
}
// Revoked 2 partitions
taskManager.RevokeTasks(new List <TopicPartition> {
new TopicPartition("topic", 2),
new TopicPartition("topic", 3),
});
Assert.AreEqual(2, taskManager.ActiveTasks.Count());
Assert.AreEqual(2, taskManager.RevokedTasks.Count());
for (int i = 0; i < 2; ++i)
{
var task = taskManager.ActiveTaskFor(new TopicPartition("topic", i));
Assert.IsNotNull(task);
Assert.AreEqual("test-app", task.ApplicationId);
Assert.IsFalse(task.CanProcess(DateTime.Now.GetMilliseconds()));
Assert.IsFalse(task.CommitNeeded);
Assert.IsFalse(task.HasStateStores);
}
var taskFailed = taskManager.ActiveTaskFor(new TopicPartition("topic", 2));
Assert.IsNull(taskFailed);
taskManager.Close();
Assert.AreEqual(0, taskManager.ActiveTasks.Count());
Assert.AreEqual(0, taskManager.RevokedTasks.Count());
}
public async Task KafkaStreamDeserializationExceptionHandlerContinueTest()
{
var _return = new List <KeyValuePair <string, string> >();
var timeout = TimeSpan.FromSeconds(10);
bool isRunningState = false;
DateTime dt = DateTime.Now;
var config = new StreamConfig <SerdesThrowException, SerdesThrowException>();
config.ApplicationId = "test";
config.BootstrapServers = "127.0.0.1";
config.PollMs = 10;
config.DeserializationExceptionHandler += (c, r, e) => ExceptionHandlerResponse.CONTINUE;
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var builder = new StreamBuilder();
builder
.Stream <string, string>("test")
.Peek((k, v) => _return.Add(KeyValuePair.Create(k, v)));
var t = builder.Build();
var stream = new KafkaStream(t, config, supplier);
stream.StateChanged += (old, @new) =>
{
if (@new.Equals(KafkaStream.State.RUNNING))
{
isRunningState = true;
}
};
await stream.StartAsync();
while (!isRunningState)
{
Thread.Sleep(250);
if (DateTime.Now > dt + timeout)
{
break;
}
}
Assert.IsTrue(isRunningState);
if (isRunningState)
{
var serdes = new StringSerDes();
dt = DateTime.Now;
producer.Produce("test",
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("k", new SerializationContext()),
Value = serdes.Serialize("test", new SerializationContext()),
Timestamp = new Confluent.Kafka.Timestamp(dt)
});
producer.Produce("test",
new Confluent.Kafka.Message <byte[], byte[]>
{
Key = serdes.Serialize("v", new SerializationContext()),
Value = serdes.Serialize("test2", new SerializationContext()),
Timestamp = new Confluent.Kafka.Timestamp(dt)
});
Thread.Sleep(1000);
var expected = new List <KeyValuePair <string, string> >();
expected.Add(KeyValuePair.Create("v", "test2"));
Assert.AreEqual(expected, _return);
}
stream.Dispose();
}
public void StreamTaskWithEXACTLY_ONCE()
{
var config = new StreamConfig <StringSerDes, StringSerDes>();
config.ApplicationId = "test-app";
config.Guarantee = ProcessingGuarantee.EXACTLY_ONCE;
var serdes = new StringSerDes();
var builder = new StreamBuilder();
builder.Stream <string, string>("topic")
.Map((k, v) => KeyValuePair.Create(k.ToUpper(), v.ToUpper()))
.To("topic2");
var topology = builder.Build();
TaskId id = new TaskId {
Id = 0, Partition = 0
};
var processorTopology = topology.Builder.BuildTopology(id);
var supplier = new SyncKafkaSupplier();
var producer = supplier.GetProducer(config.ToProducerConfig());
var consumer = supplier.GetConsumer(config.ToConsumerConfig(), null);
var part = new TopicPartition("topic", 0);
StreamTask task = new StreamTask(
"thread-0",
id,
new List <TopicPartition> {
part
},
processorTopology,
consumer,
config,
supplier,
null);
task.GroupMetadata = consumer as SyncConsumer;
task.InitializeStateStores();
task.InitializeTopology();
List <ConsumeResult <byte[], byte[]> > messages = new List <ConsumeResult <byte[], byte[]> >();
int offset = 0;
for (int i = 0; i < 5; ++i)
{
messages.Add(
new ConsumeResult <byte[], byte[]>
{
Message = new Message <byte[], byte[]>
{
Key = serdes.Serialize($"key{i + 1}", new SerializationContext()),
Value = serdes.Serialize($"value{i + 1}", new SerializationContext())
},
TopicPartitionOffset = new TopicPartitionOffset(part, offset++)
});
}
task.AddRecords(messages);
Assert.IsTrue(task.CanProcess(DateTime.Now.GetMilliseconds()));
while (task.CanProcess(DateTime.Now.GetMilliseconds()))
{
Assert.IsTrue(task.Process());
Assert.IsTrue(task.CommitNeeded);
task.Commit();
}
// CHECK IN TOPIC topic2
consumer.Subscribe("topic2");
List <ConsumeResult <byte[], byte[]> > results = new List <ConsumeResult <byte[], byte[]> >();
ConsumeResult <byte[], byte[]> result = null;
do
{
result = consumer.Consume(100);
if (result != null)
{
results.Add(result);
consumer.Commit(result);
}
} while (result != null);
Assert.AreEqual(5, results.Count);
for (int i = 0; i < 5; ++i)
{
Assert.AreEqual($"KEY{i + 1}", serdes.Deserialize(results[i].Message.Key, new SerializationContext()));
Assert.AreEqual($"VALUE{i + 1}", serdes.Deserialize(results[i].Message.Value, new SerializationContext()));
}
task.Close();
}