denis-chudov commented on code in PR #5825: URL: https://github.com/apache/ignite-3/pull/5825#discussion_r2159006189
########## modules/distribution-zones/src/integrationTest/java/org/apache/ignite/internal/rebalance/ItRebalanceByPendingAssignmentsQueueTest.java: ########## @@ -0,0 +1,565 @@ +/* + * 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.ignite.internal.rebalance; + +import static java.util.Optional.ofNullable; +import static java.util.concurrent.TimeUnit.SECONDS; +import static java.util.stream.Collectors.toList; +import static java.util.stream.Collectors.toSet; +import static org.apache.ignite.internal.TestRebalanceUtil.pendingPartitionAssignmentsKey; +import static org.apache.ignite.internal.TestRebalanceUtil.stablePartitionAssignmentsKey; +import static org.apache.ignite.internal.TestWrappers.unwrapIgniteImpl; +import static org.apache.ignite.internal.TestWrappers.unwrapTableViewInternal; +import static org.apache.ignite.internal.lang.IgniteStringFormatter.format; +import static org.apache.ignite.internal.lang.IgniteSystemProperties.colocationEnabled; +import static org.apache.ignite.internal.partitiondistribution.PendingAssignmentsCalculator.pendingAssignmentsCalculator; +import static org.apache.ignite.internal.rebalance.ItRebalanceByPendingAssignmentsQueueTest.AssignmentsRecorder.recordAssignmentsEvents; +import static org.apache.ignite.internal.testframework.IgniteTestUtils.runRace; +import static org.apache.ignite.internal.testframework.matchers.CompletableFutureMatcher.willCompleteSuccessfully; +import static org.apache.ignite.internal.testframework.matchers.CompletableFutureMatcher.willSucceedIn; +import static org.apache.ignite.internal.util.CompletableFutures.nullCompletedFuture; +import static org.awaitility.Awaitility.await; +import static org.hamcrest.MatcherAssert.assertThat; +import static org.hamcrest.Matchers.equalTo; +import static org.hamcrest.Matchers.greaterThanOrEqualTo; +import static org.hamcrest.Matchers.hasItem; +import static org.hamcrest.Matchers.hasSize; +import static org.hamcrest.Matchers.not; +import static org.junit.jupiter.api.Assertions.assertFalse; +import static org.junit.jupiter.api.Assertions.assertNotNull; +import static org.junit.jupiter.api.Assertions.assertNull; + +import java.util.Deque; +import java.util.HashSet; +import java.util.List; +import java.util.Map; +import java.util.Objects; +import java.util.Set; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.ConcurrentHashMap; +import java.util.concurrent.ConcurrentLinkedDeque; +import java.util.function.Consumer; +import java.util.function.Predicate; +import java.util.stream.Stream; +import org.apache.ignite.Ignite; +import org.apache.ignite.internal.Cluster; +import org.apache.ignite.internal.ClusterPerTestIntegrationTest; +import org.apache.ignite.internal.TestRebalanceUtil; +import org.apache.ignite.internal.TestWrappers; +import org.apache.ignite.internal.app.IgniteImpl; +import org.apache.ignite.internal.catalog.Catalog; +import org.apache.ignite.internal.catalog.CatalogManager; +import org.apache.ignite.internal.catalog.descriptors.CatalogZoneDescriptor; +import org.apache.ignite.internal.distributionzones.rebalance.RebalanceUtil; +import org.apache.ignite.internal.distributionzones.rebalance.ZoneRebalanceUtil; +import org.apache.ignite.internal.hlc.HybridTimestamp; +import org.apache.ignite.internal.lang.ByteArray; +import org.apache.ignite.internal.lang.RunnableX; +import org.apache.ignite.internal.metastorage.Entry; +import org.apache.ignite.internal.metastorage.MetaStorageManager; +import org.apache.ignite.internal.metastorage.WatchEvent; +import org.apache.ignite.internal.metastorage.WatchListener; +import org.apache.ignite.internal.partitiondistribution.Assignment; +import org.apache.ignite.internal.partitiondistribution.Assignments; +import org.apache.ignite.internal.partitiondistribution.AssignmentsQueue; +import org.apache.ignite.internal.placementdriver.ReplicaMeta; +import org.apache.ignite.internal.placementdriver.message.PlacementDriverMessagesFactory; +import org.apache.ignite.internal.placementdriver.message.StopLeaseProlongationMessage; +import org.apache.ignite.internal.raft.Peer; +import org.apache.ignite.internal.raft.PeersAndLearners; +import org.apache.ignite.internal.raft.RaftGroupEventsListener; +import org.apache.ignite.internal.raft.RaftNodeId; +import org.apache.ignite.internal.raft.server.impl.JraftServerImpl; +import org.apache.ignite.internal.replicator.PartitionGroupId; +import org.apache.ignite.internal.replicator.TablePartitionId; +import org.apache.ignite.internal.replicator.ZonePartitionId; +import org.apache.ignite.internal.table.TableViewInternal; +import org.apache.ignite.internal.table.distributed.disaster.TestDisasterRecoveryUtils; +import org.apache.ignite.raft.jraft.Node; +import org.apache.ignite.raft.jraft.RaftGroupService; +import org.apache.ignite.raft.jraft.entity.PeerId; +import org.jetbrains.annotations.Nullable; +import org.junit.jupiter.api.Test; + +/** + * Tests for rebalance - handling pending {@link AssignmentsQueue} by {@link RaftGroupEventsListener}. + */ +class ItRebalanceByPendingAssignmentsQueueTest extends ClusterPerTestIntegrationTest { + private static final int AUTO_SCALE_TIMEOUT_SECONDS = 2; + private static final String ZONE_NAME = "TEST_ZONE"; + private static final String TABLE_NAME = "TEST_TABLE"; + + @Override + protected int initialNodes() { + return 4; + } + + @Test + void testDoStableKeySwitchWhenPendingQueueIsOne() { + createZoneAndTable(4, 2); + + Set<Assignment> stableAssignments = stablePartitionAssignments(TABLE_NAME); + + // update pending assignments by remove some in stable + AssignmentsQueue expectedPendingAssignmentsQueue = assignmentsReduce(stableAssignments); + assertThat("pending queue size = 1", expectedPendingAssignmentsQueue.size(), equalTo(1)); + + recordAssignmentsEvents(cluster.aliveNode(), recorder -> { + + putPendingAssignments(cluster.aliveNode(), TABLE_NAME, expectedPendingAssignmentsQueue); + + await().untilAsserted(() -> { + Deque<AssignmentsQueue> pendingEvents = recorder.pendingAssignmentsEvents(TABLE_NAME); + Deque<Assignments> stableEvents = recorder.stableAssignmentsEvents(TABLE_NAME); + + assertThat("pending assignment events N=1", pendingEvents, hasSize(1)); + assertThat(pendingEvents.peekFirst(), equalTo(expectedPendingAssignmentsQueue)); + + assertThat("only one stable switch", stableEvents, hasSize(1)); + assertThat(stableEvents.peekLast(), equalTo(pendingEvents.peekLast().peekLast())); + }); + }); + } + + @Test + void testDoStableKeySwitchWhenPendingQueueIsGreaterThanOne() { + createZoneAndTable(4, 2); + + Set<Assignment> stableAssignments = stablePartitionAssignments(TABLE_NAME); + + // update pending assignments by promote/demote some in stable + AssignmentsQueue expectedPendingAssignmentsQueue = assignmentsPromoteDemote(stableAssignments); + assertThat("pending queue size >= 2", expectedPendingAssignmentsQueue.size(), greaterThanOrEqualTo(2)); + + recordAssignmentsEvents(cluster.aliveNode(), recorder -> { + + putPendingAssignments(cluster.aliveNode(), TABLE_NAME, expectedPendingAssignmentsQueue); + + await().untilAsserted(() -> { + Deque<AssignmentsQueue> pendingEvents = recorder.pendingAssignmentsEvents(TABLE_NAME); + Deque<Assignments> stableEvents = recorder.stableAssignmentsEvents(TABLE_NAME); + + assertThat("pending assignment events N=queue size", pendingEvents, hasSize(expectedPendingAssignmentsQueue.size())); + assertThat(pendingEvents.peekFirst(), equalTo(expectedPendingAssignmentsQueue)); + + assertThat("only one stable switch", stableEvents, hasSize(1)); + assertThat(stableEvents.peekLast(), equalTo(pendingEvents.peekLast().peekLast())); + }); + }); + } + + @Test + void testScaleUp() { + createZoneAndTable(2, 2); + + var stable0 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat(stable0.peers(), hasSize(2)); + assertThat(stable0.learners(), hasSize(0)); + + // new peer should be started + alterZone(3, 2); + var stable1 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat(stable1.peers(), hasSize(3)); + assertThat(stable1.learners(), hasSize(0)); + + // new learner should be started + alterZone(4, 2); + var stable2 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat(stable2.peers(), hasSize(3)); + assertThat(stable2.learners(), hasSize(1)); + } + + @Test + void testScaleDown() { + createZoneAndTable(4, 2); + + var stable0 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat(stable0.peers(), hasSize(3)); + assertThat(stable0.learners(), hasSize(1)); + + String leaseholder = primaryReplicaMeta(TABLE_NAME).getLeaseholder(); + + // peer goes offline + String stopNodeName = stable0.peers().stream() + .filter(p -> !p.consistentId().equals(leaseholder)) + .findFirst().map(Peer::consistentId).orElseThrow(); + cluster.stopNode(stopNodeName); + await().atMost(30, SECONDS).untilAsserted(() -> { + var stableNames = stablePartitionAssignments(TABLE_NAME).stream() + .map(Assignment::consistentId).collect(toSet()); + assertThat(stableNames, not(hasItem(stopNodeName))); + }); + + // learner promoted to peer + var stable1 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat("learner promoted to peer", stable1.peers(), hasItem(stable0.learners().iterator().next())); + assertThat(stable1.peers(), hasSize(3)); + assertThat(stable1.learners(), hasSize(0)); + } + + @Test + void testAutomaticRebalanceWhenPeersMajorityIsLostAndNoAvailableLearners() { + createZoneAndTable(4, 2); + + var stable0 = PeersAndLearners.fromAssignments(stablePartitionAssignments(TABLE_NAME)); + assertThat(stable0.peers(), hasSize(3)); + assertThat(stable0.learners(), hasSize(1)); + + String leaseholder = primaryReplicaMeta(TABLE_NAME).getLeaseholder(); + + // learners and some of majority goes offline + Stream<Peer> someQuorumNodes = stable0.peers().stream().filter(p -> !p.consistentId().equals(leaseholder)).limit(1); + Stream<Peer> learnerNodes = stable0.learners().stream(); // prevent promote learner into peer + List<String> stopNodes = Stream.concat(learnerNodes, someQuorumNodes) + .map(Peer::consistentId).collect(toList()); + assertThat(stopNodes, hasSize(2)); + + stopNodesInParallel(cluster, stopNodes); + + await().atMost(120, SECONDS).untilAsserted(() -> { + assertThat(dataNodes(TABLE_NAME), hasSize(2)); + + Assignments a = stablePartitionAssignmentsValue(TABLE_NAME); + assertNotNull(a, "stable assignments should not be null"); + assertFalse(a.fromReset(), "stable should not have fromReset flag"); + assertFalse(a.force(), "stable should not have force flag"); + + var stable1 = PeersAndLearners.fromAssignments(a.nodes()); + assertThat(stable1.peers(), hasSize(2)); + assertThat(stable1.learners(), hasSize(0)); + }); + + assertNull(pendingAssignmentsValue(TABLE_NAME), "no ongoing or planned rebalance"); + } + + @Test + void testNodeRestartDuringQueueProcessing() { + createZoneAndTable(4, 2); + + Set<Assignment> stableAssignments = stablePartitionAssignments(TABLE_NAME); + AssignmentsQueue expectedPendingAssignmentsQueue = assignmentsPromoteDemote(stableAssignments); + assertThat("pending queue size >= 2", expectedPendingAssignmentsQueue.size(), greaterThanOrEqualTo(2)); + + String leaseholder = primaryReplicaMeta(TABLE_NAME).getLeaseholder(); + String restartNode = stableAssignments.stream() + .map(Assignment::consistentId).filter(name -> !name.equals(leaseholder)).findFirst().orElseThrow(); + + putPendingAssignments(raftLeader(TABLE_NAME), TABLE_NAME, expectedPendingAssignmentsQueue); + cluster.restartNode(cluster.nodeIndex(restartNode)); + + await().atMost(120, SECONDS).untilAsserted(() -> { + assertThat(dataNodes(TABLE_NAME), hasSize(4)); Review Comment: The default failure handler is `noop` so if something fails during the recovery - we may not see it. Please add the failure handler configuration that would stop the node in case of problem: `failureHandler.type: stop` (see `StopNodeOrHaltFailureHandlerConfigurationSchema`) See also `@ConfigOverrides` for test config change. -- 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: notifications-unsubscr...@ignite.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
