dajac commented on code in PR #23732: URL: https://github.com/apache/kafka/pull/23732#discussion_r4249031985
########## group-coordinator/src/test/java/org/apache/kafka/coordinator/group/assignor/uniform2/GroupModelTest.java: ########## @@ -0,0 +1,310 @@ +/* + * 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.coordinator.group.assignor.uniform2; + +import org.apache.kafka.common.Uuid; +import org.apache.kafka.coordinator.common.runtime.CoordinatorMetadataImage; +import org.apache.kafka.coordinator.common.runtime.MetadataImageBuilder; +import org.apache.kafka.coordinator.group.api.assignor.SubscribedTopicDescriber; +import org.apache.kafka.coordinator.group.api.assignor.SubscriptionType; +import org.apache.kafka.coordinator.group.modern.SubscribedTopicDescriberImpl; +import org.apache.kafka.coordinator.group.modern.TopicIds; + +import org.junit.jupiter.api.Test; + +import java.util.Arrays; +import java.util.LinkedHashSet; +import java.util.List; +import java.util.Set; +import java.util.stream.IntStream; + +import static org.junit.jupiter.api.Assertions.assertEquals; + +/** + * Tests of {@link GroupModel}: the numbering of the members and the topics, and the cohorts. + * + * <p>The topic ids sort as T1 < T2 < T3, so that when all of them are subscribed T1 is + * topic 0, T2 is topic 1 and T3 is topic 2; the member ids sort as A < B < C < D, so + * that A is member 0, B is member 1, and so on. The members are put in the specs in another + * order, and the subscriptions list their topics in another order, so that the numbers cannot + * come from the iteration order of the spec. + */ +public class GroupModelTest { + private static final Uuid T1 = new Uuid(1L, 1L); + private static final Uuid T2 = new Uuid(2L, 1L); + private static final Uuid T3 = new Uuid(3L, 1L); + private static final Uuid MISSING = new Uuid(9L, 1L); + + private static final SubscribedTopicDescriber DESCRIBER = new TopicsFixture() + .withTopic(T1, 2) + .withTopic(T2, 5) + .withTopic(T3, 7) + .build(); + + /** + * Member ids are compared as strings, so m10 comes between m1 and m2. + */ + @Test + public void testMembersAreNumberedInMemberIdOrder() { + var spec = new GroupSpecFixture() + .withMember("m2", Set.of(T1)) + .withMember("m10", Set.of(T1)) + .withMember("m1", Set.of(T1)); + + var group = new GroupModel(spec.build(), DESCRIBER); + + assertEquals(List.of("m1", "m10", "m2"), memberIds(group)); + } + + /** + * A, read first, meets T3 before T1, and B then meets T2; the topics are still numbered in + * topic id order, with their partition counts. + */ + @Test + public void testTopicsAreNumberedInTopicIdOrder() { + var spec = new GroupSpecFixture() + .withMember("B", Set.of(T2)) + .withMember("A", ordered(T3, T1)); + + var group = new GroupModel(spec.build(), DESCRIBER); + + assertEquals(List.of(T1, T2, T3), topicIds(group)); + assertEquals(List.of(0, 1, 2), topicIndexes(group, T1, T2, T3)); + assertEquals(List.of(2, 5, 7), partitionCounts(group)); + assertEquals(List.of(List.of(0, 2), List.of(1)), topicsOfCohorts(group)); + } + + /** + * A subscribes to T1 and T2, B to T2 and T3, C to T1 and T2 like A, and D to T3. Read in + * member order, A starts cohort 0, B cohort 1, C joins cohort 0 although it lists its topics + * in another order, and D starts cohort 2. T1 then has 2 subscribers (A and C), T2 has 3 (A, + * B and C) and T3 has 2 (B and D). + */ + @Test + public void testCohorts() { + var spec = new GroupSpecFixture() + .withMember("D", Set.of(T3)) + .withMember("C", ordered(T2, T1)) + .withMember("B", ordered(T3, T2)) + .withMember("A", ordered(T1, T2)); + + var group = new GroupModel(spec.build(), DESCRIBER); + + assertEquals(List.of(0, 1, 0, 2), cohorts(group)); + assertEquals(List.of(List.of(0, 2), List.of(1), List.of(3)), membersOfCohorts(group)); + assertEquals(List.of(List.of(0, 1), List.of(1, 2), List.of(2)), topicsOfCohorts(group)); + assertEquals(List.of(List.of(0), List.of(0, 1), List.of(1, 2)), cohortsOfTopics(group)); + assertEquals(List.of(2, 3, 2), subscriberCounts(group)); + assertEquals(List.of(List.of(0, 1), List.of(1, 2), List.of(0, 1), List.of(2)), subscribedTopics(group)); + } + + /** + * The describer gives -1 partitions for a topic that does not exist, which is then dropped + * from the subscriptions: A, subscribing to T1 and the missing topic, and B, subscribing to T1, + * share cohort 0, and C, subscribing only to the missing topic, has a cohort without topics. + */ + @Test + public void testTopicsThatDoNotExistAreIgnored() { + var spec = new GroupSpecFixture() + .withMember("C", Set.of(MISSING)) + .withMember("B", Set.of(T1)) + .withMember("A", ordered(MISSING, T1)); + + var group = new GroupModel(spec.build(), DESCRIBER); + + assertEquals(List.of(T1), topicIds(group)); + assertEquals(List.of(0, -1), topicIndexes(group, T1, MISSING)); + assertEquals(List.of(2), subscriberCounts(group)); + assertEquals(List.of(List.of(0, 1), List.of(2)), membersOfCohorts(group)); + assertEquals(List.of(List.of(0), List.of()), topicsOfCohorts(group)); + assertEquals(List.of(List.of(0)), cohortsOfTopics(group)); + assertEquals(List.of(List.of(0), List.of(0), List.of()), subscribedTopics(group)); + } + + /** + * A, B and C subscribe to T1 and T2, listing them in different orders: the group is + * homogeneous, and makes a single cohort with every member and topic. + */ + @Test + public void testHomogeneousGroup() { + var spec = new GroupSpecFixture() + .withMember("C", ordered(T2, T1)) + .withMember("B", ordered(T1, T2)) + .withMember("A", ordered(T2, T1)) + .build(); + assertEquals(SubscriptionType.HOMOGENEOUS, spec.subscriptionType()); + + var group = new GroupModel(spec, DESCRIBER); + + assertEquals(List.of("A", "B", "C"), memberIds(group)); + assertEquals(List.of(T1, T2), topicIds(group)); + assertEquals(List.of(3, 3), subscriberCounts(group)); + assertEquals(List.of(0, 0, 0), cohorts(group)); + assertEquals(List.of(List.of(0, 1, 2)), membersOfCohorts(group)); + assertEquals(List.of(List.of(0, 1)), topicsOfCohorts(group)); + assertEquals(List.of(List.of(0), List.of(0)), cohortsOfTopics(group)); + assertEquals(List.of(List.of(0, 1), List.of(0, 1), List.of(0, 1)), subscribedTopics(group)); + } + + /** + * The coordinator gives the subscriptions as {@link TopicIds}, resolving topic names against + * the metadata image when iterated; an unknown name is skipped, although it counts in their + * size. A subscribes to t1, t3 and an unknown topic, B to t1 and t2, and C to t1 and t3: A and + * C have the same topics, so they share cohort 0. Three members subscribing to all the topics + * and the unknown one make a homogeneous group, a single cohort with the three topics. + */ + @Test + public void testSubscriptionsAsTopicIds() { Review Comment: The size of `TopicIds` is awkward. The size is based on the underlying names but it may differ from the resolved one. We don't really need this test though. Let me remove it. -- 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]
