Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Commit offsets at specified intervals #1405

Merged
merged 3 commits into from
Nov 9, 2021
Merged

Commit offsets at specified intervals #1405

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ea89ed0
Commit offset at a specified interval
pierDipi Nov 3, 2021
d2776e4
Add reset logic to handle long large offsets
pierDipi Nov 8, 2021
b582863
Use final
pierDipi Nov 9, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
74 changes: 25 additions & 49 deletions ...ative/eventing/kafka/broker/dispatcher/impl/consumer/UnorderedOffsetManagerBenchmark.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@
import io.vertx.core.AsyncResult;
import io.vertx.core.Future;
import io.vertx.core.Handler;
import io.vertx.core.Vertx;
import io.vertx.kafka.client.common.PartitionInfo;
import io.vertx.kafka.client.common.TopicPartition;
import io.vertx.kafka.client.consumer.KafkaConsumerRecord;
Expand All @@ -27,11 +28,6 @@
import io.vertx.kafka.client.consumer.OffsetAndMetadata;
import io.vertx.kafka.client.consumer.OffsetAndTimestamp;
import io.vertx.kafka.client.consumer.impl.KafkaConsumerRecordImpl;
import java.time.Duration;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.regex.Pattern;
import org.apache.kafka.clients.consumer.Consumer;
import org.apache.kafka.clients.consumer.ConsumerRecord;
import org.openjdk.jmh.annotations.Benchmark;
Expand All @@ -41,6 +37,12 @@
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.infra.Blackhole;

import java.time.Duration;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.regex.Pattern;

public class UnorderedOffsetManagerBenchmark {

@State(Scope.Thread)
Expand Down Expand Up @@ -71,96 +73,70 @@ public void doSetup() {

@Benchmark
public void benchmarkReverseOrder(RecordsState recordsState, Blackhole blackhole) {
UnorderedOffsetManager offsetManager = new UnorderedOffsetManager(new MockKafkaConsumer(), null);
OffsetManager offsetManager = new OffsetManager(Vertx.vertx(), new MockKafkaConsumer(), null, 10000L);
pierDipi marked this conversation as resolved.
Show resolved Hide resolved

int partitions = 100;
for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.recordReceived(recordsState.records[partition][0])
);
offsetManager.recordReceived(recordsState.records[partition][0]);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

are they directly consumed?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I updated the code to make it compile but I'm inclined to drop these OffsetManager benchmarks since they are not really useful to catch perf regressions as they are.

}

for (int offset = 9_999; offset > 0; offset--) {
for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.recordReceived(recordsState.records[partition][offset])
);
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset])
);
offsetManager.recordReceived(recordsState.records[partition][offset]);
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset]);
}
}

for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[partition][0])
);
offsetManager.successfullySentToSubscriber(recordsState.records[partition][0]);
}
}

@Benchmark
public void benchmarkOrdered(RecordsState recordsState, Blackhole blackhole) {
UnorderedOffsetManager offsetManager = new UnorderedOffsetManager(new MockKafkaConsumer(), null);
OffsetManager offsetManager = new OffsetManager(Vertx.vertx(), new MockKafkaConsumer(), null, 10000L);
int partitions = 100;

for (int offset = 0; offset < 10_000; offset++) {
for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.recordReceived(recordsState.records[partition][offset])
);
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset])
);
offsetManager.recordReceived(recordsState.records[partition][offset]);
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset]);
}
}
}

@Benchmark
public void benchmarkRealisticCase(RecordsState recordsState, Blackhole blackhole) {
UnorderedOffsetManager offsetManager = new UnorderedOffsetManager(new MockKafkaConsumer(), null);
OffsetManager offsetManager = new OffsetManager(Vertx.vertx(), new MockKafkaConsumer(), null, 10000L);
int partitions = 10;

for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.recordReceived(recordsState.records[partition][0])
);
offsetManager.recordReceived(recordsState.records[partition][0]);
}

for (int partition = 0; partition < partitions; partition++) {
for (int offset : new int[] {5, 2, 0, 7, 1, 3, 4, 6}) {
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset])
);
for (int offset : new int[]{5, 2, 0, 7, 1, 3, 4, 6}) {
offsetManager.successfullySentToSubscriber(recordsState.records[partition][offset]);
}
}
}

@Benchmark
public void benchmarkMixedABit(RecordsState recordsState, Blackhole blackhole) {
UnorderedOffsetManager offsetManager = new UnorderedOffsetManager(new MockKafkaConsumer(), null);
OffsetManager offsetManager = new OffsetManager(Vertx.vertx(), new MockKafkaConsumer(), null, 10000L);
int partitions = 4;

for (int partition = 0; partition < partitions; partition++) {
blackhole.consume(
offsetManager.recordReceived(recordsState.records[partition][0])
);
offsetManager.recordReceived(recordsState.records[partition][0]);
}

for (int i = 0; i < 120; i++) {
// This will commit in the following order:
// 1 0 3 2 5 4 ...
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[2][i % 2 == 0 ? i + 1 : i - 1])
);
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[1][i % 2 == 0 ? i + 1 : i - 1])
);
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[0][i % 2 == 0 ? i + 1 : i - 1])
);
blackhole.consume(
offsetManager.successfullySentToSubscriber(recordsState.records[3][i % 2 == 0 ? i + 1 : i - 1])
);
offsetManager.successfullySentToSubscriber(recordsState.records[2][i % 2 == 0 ? i + 1 : i - 1]);
offsetManager.successfullySentToSubscriber(recordsState.records[1][i % 2 == 0 ? i + 1 : i - 1]);
offsetManager.successfullySentToSubscriber(recordsState.records[0][i % 2 == 0 ? i + 1 : i - 1]);
offsetManager.successfullySentToSubscriber(recordsState.records[3][i % 2 == 0 ? i + 1 : i - 1]);
}
}

Expand Down
10 changes: 5 additions & 5 deletions .../src/main/java/dev/knative/eventing/kafka/broker/dispatcher/RecordDispatcherListener.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,34 +28,34 @@ public interface RecordDispatcherListener {
*
* @param record record received.
*/
Future<Void> recordReceived(KafkaConsumerRecord<?, ?> record);
void recordReceived(KafkaConsumerRecord<?, ?> record);

/**
* The given record cannot be delivered to dead letter sink.
*
* @param record record undeliverable to dead letter sink.
* @param ex exception occurred.
*/
Future<Void> failedToSendToDeadLetterSink(KafkaConsumerRecord<?, ?> record, Throwable ex);
void failedToSendToDeadLetterSink(KafkaConsumerRecord<?, ?> record, Throwable ex);

/**
* The given event doesn't pass the filter.
*
* @param record record discarded.
*/
Future<Void> recordDiscarded(KafkaConsumerRecord<?, ?> record);
void recordDiscarded(KafkaConsumerRecord<?, ?> record);

/**
* The given record has been successfully sent to subscriber.
*
* @param record record sent to subscriber.
*/
Future<Void> successfullySentToSubscriber(KafkaConsumerRecord<?, ?> record);
void successfullySentToSubscriber(KafkaConsumerRecord<?, ?> record);

/**
* The given record has been successfully sent to dead letter sink.
*
* @param record record sent to dead letter sink.
*/
Future<Void> successfullySentToDeadLetterSink(KafkaConsumerRecord<?, ?> record);
void successfullySentToDeadLetterSink(KafkaConsumerRecord<?, ?> record);
}
35 changes: 17 additions & 18 deletions ...src/main/java/dev/knative/eventing/kafka/broker/dispatcher/impl/RecordDispatcherImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,8 +28,10 @@
import io.vertx.core.Vertx;
import io.vertx.kafka.client.common.tracing.ConsumerTracer;
import io.vertx.kafka.client.consumer.KafkaConsumerRecord;

import java.util.Objects;
import java.util.function.Function;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

Expand Down Expand Up @@ -139,16 +141,13 @@ private void onRecordReceived(final KafkaConsumerRecord<Object, CloudEvent> reco
}
}

recordDispatcherListener.recordReceived(record)
.onSuccess(v -> {
// Execute filtering
if (filter.test(record.value())) {
onFilterMatching(record, finalProm);
} else {
onFilterNotMatching(record, finalProm);
}
})
.onFailure(finalProm::fail); // This should really never happen
recordDispatcherListener.recordReceived(record);
// Execute filtering
if (filter.test(record.value())) {
onFilterMatching(record, finalProm);
} else {
onFilterNotMatching(record, finalProm);
}
}

private void onFilterMatching(final KafkaConsumerRecord<Object, CloudEvent> record, final Promise<Void> finalProm) {
Expand All @@ -161,15 +160,15 @@ private void onFilterMatching(final KafkaConsumerRecord<Object, CloudEvent> reco
private void onFilterNotMatching(final KafkaConsumerRecord<Object, CloudEvent> record,
final Promise<Void> finalProm) {
logDebug("Record doesn't match filtering", record);
recordDispatcherListener.recordDiscarded(record)
.onComplete(finalProm);
recordDispatcherListener.recordDiscarded(record);
finalProm.complete();
}

private void onSubscriberSuccess(final KafkaConsumerRecord<Object, CloudEvent> record,
final Promise<Void> finalProm) {
logDebug("Successfully sent event to subscriber", record);
recordDispatcherListener.successfullySentToSubscriber(record)
.onComplete(finalProm);
recordDispatcherListener.successfullySentToSubscriber(record);
finalProm.complete();
}

private void onSubscriberFailure(final KafkaConsumerRecord<Object, CloudEvent> record,
Expand All @@ -182,15 +181,15 @@ private void onSubscriberFailure(final KafkaConsumerRecord<Object, CloudEvent> r
private void onDeadLetterSinkSuccess(final KafkaConsumerRecord<Object, CloudEvent> record,
final Promise<Void> finalProm) {
logDebug("Successfully sent event to the dead letter sink", record);
recordDispatcherListener.successfullySentToDeadLetterSink(record)
.onComplete(finalProm);
recordDispatcherListener.successfullySentToDeadLetterSink(record);
finalProm.complete();
}


private void onDeadLetterSinkFailure(final KafkaConsumerRecord<Object, CloudEvent> record, final Throwable exception,
final Promise<Void> finalProm) {
recordDispatcherListener.failedToSendToDeadLetterSink(record, exception)
.onComplete(finalProm);
recordDispatcherListener.failedToSendToDeadLetterSink(record, exception);
finalProm.complete();
}

private static Function<KafkaConsumerRecord<Object, CloudEvent>, Future<Void>> composeSenderAndSinkHandler(
Expand Down
Loading