AndrewJSchofield commented on code in PR #21335:
URL: https://github.com/apache/kafka/pull/21335#discussion_r2718751262
##########
clients/src/main/java/org/apache/kafka/clients/admin/internals/AdminApiDriver.java:
##########
@@ -115,8 +115,13 @@ public AdminApiDriver(
// metadata. For all cached keys, they can proceed straight to the
fulfillment map.
// Note that the cache is only used on the initial calls, and any
errors that result
// in additional lookups use the full set of lookup keys.
- retryLookup(future.uncachedLookupKeys());
Review Comment:
Can't you remove `uncachedLookupKeys` entirely?
##########
clients/clients-integration-tests/src/test/java/org/apache/kafka/clients/admin/ConcurrentListOffsetsRequestTest.java:
##########
@@ -0,0 +1,204 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.kafka.clients.admin;
+
+import org.apache.kafka.clients.CommonClientConfigs;
+import org.apache.kafka.clients.DefaultHostResolver;
+import org.apache.kafka.clients.NetworkClient;
+import org.apache.kafka.clients.admin.internals.PartitionLeaderCache;
+import org.apache.kafka.common.IsolationLevel;
+import org.apache.kafka.common.TopicPartition;
+import org.apache.kafka.common.errors.TimeoutException;
+import org.apache.kafka.common.test.ClusterInstance;
+import org.apache.kafka.common.test.api.ClusterTest;
+import org.apache.kafka.common.test.api.ClusterTestDefaults;
+import org.apache.kafka.common.test.api.Type;
+import org.apache.kafka.common.utils.Utils;
+
+import org.junit.jupiter.api.AfterEach;
+import org.junit.jupiter.api.BeforeEach;
+
+import java.lang.reflect.Field;
+import java.net.InetAddress;
+import java.net.UnknownHostException;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.CountDownLatch;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.concurrent.atomic.AtomicInteger;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+import static org.junit.jupiter.api.Assertions.assertInstanceOf;
+import static org.junit.jupiter.api.Assertions.assertThrows;
+
+@ClusterTestDefaults(
+ types = {Type.KRAFT},
+ brokers = 3
+)
+public class ConcurrentListOffsetsRequestTest {
+ private static final String TOPIC = "topic";
+ private static final short REPLICAS = 1;
+ private static final int PARTITION = 2;
+ private static final int TIMEOUT = 1000;
+ private final ClusterInstance clusterInstance;
+ private Admin adminClient;
+ private NetworkClient networkClient;
+ private final AtomicBoolean injectHostResolverError = new
AtomicBoolean(false);
+
+ ConcurrentListOffsetsRequestTest(ClusterInstance clusterInstance) {
+ this.clusterInstance = clusterInstance;
+ }
+
+ @BeforeEach
+ public void setup() throws Exception {
+ clusterInstance.waitForReadyBrokers();
+ clusterInstance.createTopic(TOPIC, PARTITION, REPLICAS);
+ Map<String, Object> props = Map.of(
+ "default.api.timeout.ms", TIMEOUT,
+ "request.timeout.ms", TIMEOUT,
+ CommonClientConfigs.BOOTSTRAP_SERVERS_CONFIG,
clusterInstance.bootstrapServers());
+ adminClient = KafkaAdminClient.createInternal(new
AdminClientConfig(clusterInstance.setClientSaslConfig(props), true),
+ null, new TestHostResolver());
+
+ Field clientField = KafkaAdminClient.class.getDeclaredField("client");
Review Comment:
I would prefer `TestUtils.fieldValue` here.
##########
clients/clients-integration-tests/src/test/java/org/apache/kafka/clients/admin/ConcurrentListOffsetsRequestTest.java:
##########
@@ -0,0 +1,204 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.kafka.clients.admin;
+
+import org.apache.kafka.clients.CommonClientConfigs;
+import org.apache.kafka.clients.DefaultHostResolver;
+import org.apache.kafka.clients.NetworkClient;
+import org.apache.kafka.clients.admin.internals.PartitionLeaderCache;
+import org.apache.kafka.common.IsolationLevel;
+import org.apache.kafka.common.TopicPartition;
+import org.apache.kafka.common.errors.TimeoutException;
+import org.apache.kafka.common.test.ClusterInstance;
+import org.apache.kafka.common.test.api.ClusterTest;
+import org.apache.kafka.common.test.api.ClusterTestDefaults;
+import org.apache.kafka.common.test.api.Type;
+import org.apache.kafka.common.utils.Utils;
+
+import org.junit.jupiter.api.AfterEach;
+import org.junit.jupiter.api.BeforeEach;
+
+import java.lang.reflect.Field;
+import java.net.InetAddress;
+import java.net.UnknownHostException;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.CountDownLatch;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeUnit;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.concurrent.atomic.AtomicInteger;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+import static org.junit.jupiter.api.Assertions.assertInstanceOf;
+import static org.junit.jupiter.api.Assertions.assertThrows;
+
+@ClusterTestDefaults(
+ types = {Type.KRAFT},
+ brokers = 3
+)
+public class ConcurrentListOffsetsRequestTest {
+ private static final String TOPIC = "topic";
+ private static final short REPLICAS = 1;
+ private static final int PARTITION = 2;
+ private static final int TIMEOUT = 1000;
+ private final ClusterInstance clusterInstance;
+ private Admin adminClient;
+ private NetworkClient networkClient;
+ private final AtomicBoolean injectHostResolverError = new
AtomicBoolean(false);
+
+ ConcurrentListOffsetsRequestTest(ClusterInstance clusterInstance) {
+ this.clusterInstance = clusterInstance;
+ }
+
+ @BeforeEach
+ public void setup() throws Exception {
+ clusterInstance.waitForReadyBrokers();
+ clusterInstance.createTopic(TOPIC, PARTITION, REPLICAS);
+ Map<String, Object> props = Map.of(
+ "default.api.timeout.ms", TIMEOUT,
+ "request.timeout.ms", TIMEOUT,
+ CommonClientConfigs.BOOTSTRAP_SERVERS_CONFIG,
clusterInstance.bootstrapServers());
+ adminClient = KafkaAdminClient.createInternal(new
AdminClientConfig(clusterInstance.setClientSaslConfig(props), true),
+ null, new TestHostResolver());
+
+ Field clientField = KafkaAdminClient.class.getDeclaredField("client");
+ clientField.setAccessible(true);
+ networkClient = (NetworkClient) clientField.get(adminClient);
+ }
+
+ @AfterEach
+ public void teardown() {
+ Utils.closeQuietly(adminClient, "ListOffsetsAdminClient");
+ }
+
+ @ClusterTest
+ public void correctlyHandleConcurrentModificationOfPartitionLeaderCache()
throws Exception {
+ // making one request to prepopulate the partition leader cache so we
have something to delete later
+ listAllOffsets().all().get(TIMEOUT * 2, TimeUnit.SECONDS);
+
+ final CountDownLatch invalidationLatch = new CountDownLatch(1);
+ // Replacing the partition leader cache in order to be able to
synchronize the calls so that they happen in the right order to reproduce the
issue
+ TestPartitionLeaderCache testPartitionLeaderCache =
replacePartitionLeaderCache(invalidationLatch);
+
+ // closing the connection to the first node. not using
clusterInstance.shutdownBroker to reduce flakiness
+
networkClient.close(testPartitionLeaderCache.get(getTopicPartitions()).values().iterator().next().toString());
+ // as next call with try to resolve the host for the closed node, it's
time to let it fail, which will lead to cache invalidation
+ injectHostResolverError.set(true);
+
+ // making another request(this request will face the host resolver
error and remove the node from the cache)
+ ListOffsetsResult failInducingResult = listAllOffsets();
+ // waiting until we get to the invalidation
+ invalidationLatch.await();
+ // making another request. at this point the fail inducing request is
waiting for this one before it deletes the keys associated with the node
+ // the TestPartitionLeaderCache class synchronizes the calls to mimic
the race condition
+ ListOffsetsResult failingResult = listAllOffsets();
+
+ // verifying that we correctly declined the call
+ ExecutionException executionException =
assertThrows(ExecutionException.class, () ->
failInducingResult.all().get(TIMEOUT * 2, TimeUnit.MILLISECONDS));
+ assertInstanceOf(TimeoutException.class,
executionException.getCause());
+
+ // verifying that we correctly declined the call
+ executionException = assertThrows(ExecutionException.class, () ->
failingResult.all().get(TIMEOUT * 2, TimeUnit.MILLISECONDS));
+ assertInstanceOf(TimeoutException.class,
executionException.getCause());
+ }
+
+ private TestPartitionLeaderCache
replacePartitionLeaderCache(CountDownLatch invalidationLatch) throws Exception {
+ Field partitionLeaderCacheField =
KafkaAdminClient.class.getDeclaredField("partitionLeaderCache");
+ partitionLeaderCacheField.setAccessible(true);
+ PartitionLeaderCache oldPartitionLeaderCache = (PartitionLeaderCache)
partitionLeaderCacheField.get(adminClient);
+
+ TestPartitionLeaderCache partitionLeaderCache = new
TestPartitionLeaderCache(oldPartitionLeaderCache.get(getTopicPartitions()),
invalidationLatch);
+ partitionLeaderCacheField.set(adminClient, partitionLeaderCache);
+ return partitionLeaderCache;
+ }
+
+ private ListOffsetsResult listAllOffsets() {
+ List<TopicPartition> partitions = getTopicPartitions();
+
+ Map<TopicPartition, OffsetSpec> offsetSpecMap =
partitions.stream().collect(Collectors.toMap(Function.identity(), tp ->
OffsetSpec.latest()));
+ return adminClient.listOffsets(offsetSpecMap, new
ListOffsetsOptions(IsolationLevel.READ_UNCOMMITTED));
+ }
+
+ private List<TopicPartition> getTopicPartitions() {
+ List<TopicPartition> partitions = new ArrayList<>();
+ for (int i = 0; i < PARTITION; i++) {
+ partitions.add(new TopicPartition(TOPIC, i));
+ }
+ return partitions;
+ }
+
+ private static class TestPartitionLeaderCache extends PartitionLeaderCache
{
+
+ private final AtomicInteger getCounter = new AtomicInteger(0);
+ private final CountDownLatch invalidationLatch;
+ private final CountDownLatch newRequestCheckLatch = new
CountDownLatch(1);
+ private final CountDownLatch removeCompleteLatch = new
CountDownLatch(1);
+
+ public TestPartitionLeaderCache(Map<TopicPartition, Integer>
oldPartitionLeaderCache, final CountDownLatch invalidationLatch) {
+ put(oldPartitionLeaderCache);
+ this.invalidationLatch = invalidationLatch;
+ }
+
+ @Override
+ public Map<TopicPartition, Integer> get(Collection<TopicPartition>
keys) {
+ Map<TopicPartition, Integer> result = super.get(keys);
+ // waiting for the third call: first one was to close the network
connection, second one was from the request that invalidates the cache
+ if (getCounter.incrementAndGet() == 3) {
+ newRequestCheckLatch.countDown();
+ try {
+ // letting the remove method proceed and actually remove
the data
+ removeCompleteLatch.await();
+ } catch (InterruptedException e) {
+ throw new RuntimeException(e);
+ }
+ }
+
+ return result;
+ }
+
+ @Override
+ public void remove(Collection<TopicPartition> keys) {
+ try {
+ // letting the caller know that we've reached the invalidation
step, and it's time to send the second request
+ invalidationLatch.countDown();
+ // waiting for the second request to reach get
+ newRequestCheckLatch.await();
+ } catch (InterruptedException e) {
+ throw new RuntimeException(e);
+ }
+ super.remove(keys);
+ // once the value removed, we are letting the get method proceed
and return the value
+ removeCompleteLatch.countDown();
+ }
+ }
+
+ private class TestHostResolver extends DefaultHostResolver {
+
+ @Override
+ public InetAddress[] resolve(String host) throws UnknownHostException {
+ System.out.println("RESOLVE: " + host);
Review Comment:
Please remove the println.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
