Skip to content

[Issue 5597][pulsar-client-java] retry when getPartitionedTopicMetadata failed #5603

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

Merged
sijie merged 2 commits into from
Nov 20, 2019
Merged

[Issue 5597][pulsar-client-java] retry when getPartitionedTopicMetadata failed #5603

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
43303d4
Add retry when create producer async
ltamber Nov 11, 2019
4408774
fix: unit test
ltamber Nov 13, 2019
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
63 changes: 63 additions & 0 deletions pulsar-broker/src/test/java/org/apache/pulsar/stats/client/PulsarBrokerStatsClientTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,8 @@
import org.apache.pulsar.client.api.Message;
import org.apache.pulsar.client.api.Producer;
import org.apache.pulsar.client.api.ProducerConsumerBase;
import org.apache.pulsar.client.api.PulsarClient;
import org.apache.pulsar.client.api.PulsarClientException;
import org.apache.pulsar.common.policies.data.PersistentTopicInternalStats;
import org.apache.pulsar.common.policies.data.PersistentTopicInternalStats.CursorStats;
import org.slf4j.Logger;
Expand All @@ -47,6 +49,7 @@
import static org.mockito.Mockito.spy;
import static org.testng.Assert.assertEquals;
import static org.testng.Assert.assertTrue;
import static org.testng.Assert.fail;

public class PulsarBrokerStatsClientTest extends ProducerConsumerBase {

Expand Down Expand Up @@ -132,5 +135,65 @@ public void testTopicInternalStats() throws Exception {
log.info("-- Exiting {} test --", methodName);
}

@Test
public void testGetPartitionedTopicMetaData() throws Exception {
log.info("-- Starting {} test --", methodName);

final String topicName = "persistent://my-property/my-ns/my-topic1";
final String subscriptionName = "my-subscriber-name";



try {
String url = "http://localhost:51000,localhost:" + BROKER_WEBSERVICE_PORT;
if (isTcpLookup) {
url = "pulsar://localhost:51000,localhost:" + BROKER_PORT;
}
PulsarClient client = newPulsarClient(url, 0);

Consumer<byte[]> consumer = client.newConsumer().topic(topicName).subscriptionName(subscriptionName)
.acknowledgmentGroupTime(0, TimeUnit.SECONDS).subscribe();
Producer<byte[]> producer = client.newProducer().topic(topicName).create();

consumer.close();
producer.close();
client.close();
} catch (PulsarClientException pce) {
log.error("create producer or consumer error: ", pce);
fail();
}

log.info("-- Exiting {} test --", methodName);
}

@Test (timeOut = 4000)
public void testGetPartitionedTopicDataTimeout() {
log.info("-- Starting {} test --", methodName);

final String topicName = "persistent://my-property/my-ns/my-topic1";

String url = "http://localhost:51000,localhost:51001";
if (isTcpLookup) {
url = "pulsar://localhost:51000,localhost:51001";
}

PulsarClient client;
try {
client = PulsarClient.builder()
.serviceUrl(url)
.statsInterval(0, TimeUnit.SECONDS)
.operationTimeout(3, TimeUnit.SECONDS)
.build();

Producer<byte[]> producer = client.newProducer().topic(topicName).create();

fail();
} catch (PulsarClientException pce) {

@sijie sijie Nov 11, 2019

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

assert the exception time.

log.error("create producer error: ", pce);
}

log.info("-- Exiting {} test --", methodName);
}

private static final Logger log = LoggerFactory.getLogger(PulsarBrokerStatsClientTest.class);
}
6 changes: 4 additions & 2 deletions pulsar-client/src/main/java/org/apache/pulsar/client/impl/HttpClient.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@
import java.io.Closeable;
import java.io.IOException;
import java.net.HttpURLConnection;
import java.net.URI;
import java.net.URL;
import java.util.Map;
import java.util.Map.Entry;
Expand Down Expand Up @@ -127,8 +128,9 @@ public void close() throws IOException {
public <T> CompletableFuture<T> get(String path, Class<T> clazz) {
final CompletableFuture<T> future = new CompletableFuture<>();
try {
String requestUrl = new URL(serviceNameResolver.resolveHostUri().toURL(), path).toString();
String remoteHostName = serviceNameResolver.resolveHostUri().getHost();
Comment on lines -130 to -131

@ltamber ltamber Nov 13, 2019

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

@sijie maybe a bug here, the host in requestUrl not equal with remoteHostName if multi-broker service url provided, and will step forward two service url every request. please take a look.

@sijie sijie Nov 14, 2019

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

sorry I don't quite get this. Can you show me an example?

@ltamber ltamber Nov 15, 2019

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

the PulsarServiceNameResolver.resolveHostUri() called twice(every time it will return the next address) inside HttpClient.get() ,
if service url is http://ip1:port1,ip2:port2, we assume requestUrl is ip1:port1, the remoteHostName will be ip2, and the next time call HttpClient.get() the requestUrl is still ip1:port1, the remoteHostName is ip2. so HttpClient.get() will only request ip1:port1

@sijie sijie Nov 15, 2019

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

oh I see. yes. the original code has a bug there.

URI hostUri = serviceNameResolver.resolveHostUri();
String requestUrl = new URL(hostUri.toURL(), path).toString();
String remoteHostName = hostUri.getHost();
AuthenticationDataProvider authData = authentication.getAuthData(remoteHostName);

CompletableFuture<Map<String, String>> authFuture = new CompletableFuture<>();
Expand Down
32 changes: 30 additions & 2 deletions pulsar-client/src/main/java/org/apache/pulsar/client/impl/PulsarClientImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -641,17 +641,45 @@ public CompletableFuture<Integer> getNumberOfPartitions(String topic) {

public CompletableFuture<PartitionedTopicMetadata> getPartitionedTopicMetadata(String topic) {

CompletableFuture<PartitionedTopicMetadata> metadataFuture;
CompletableFuture<PartitionedTopicMetadata> metadataFuture = new CompletableFuture<>();

try {
TopicName topicName = TopicName.get(topic);
metadataFuture = lookup.getPartitionedTopicMetadata(topicName);
AtomicLong opTimeoutMs = new AtomicLong(conf.getOperationTimeoutMs());
Backoff backoff = new BackoffBuilder()

@sijie sijie Nov 11, 2019

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

Not direct related to this change. but we should consider adding backoff for other places as well.

.setInitialTime(100, TimeUnit.MILLISECONDS)
.setMandatoryStop(opTimeoutMs.get() * 2, TimeUnit.MILLISECONDS)
.setMax(0, TimeUnit.MILLISECONDS)
.useUserConfiguredIntervals(conf.getDefaultBackoffIntervalNanos(),
conf.getMaxBackoffIntervalNanos())
.create();
getPartitionedTopicMetadata(topicName, backoff, opTimeoutMs, metadataFuture);
} catch (IllegalArgumentException e) {
return FutureUtil.failedFuture(new PulsarClientException.InvalidConfigurationException(e.getMessage()));
}
return metadataFuture;
}

private void getPartitionedTopicMetadata(TopicName topicName,
Backoff backoff,
AtomicLong remainingTime,
CompletableFuture<PartitionedTopicMetadata> future) {
lookup.getPartitionedTopicMetadata(topicName).thenAccept(future::complete).exceptionally(e -> {
long nextDelay = Math.min(backoff.next(), remainingTime.get());
if (nextDelay <= 0) {
future.completeExceptionally(new PulsarClientException
.TimeoutException("Could not getPartitionedTopicMetadata within configured timeout."));
return null;
}

timer.newTimeout( task -> {
remainingTime.addAndGet(-nextDelay);
getPartitionedTopicMetadata(topicName, backoff, remainingTime, future);
}, nextDelay, TimeUnit.MILLISECONDS);
return null;
});
}

@Override
public CompletableFuture<List<String>> getPartitionsForTopic(String topic) {
return getPartitionedTopicMetadata(topic).thenApply(metadata -> {
Expand Down