ascherbakoff commented on code in PR #7731:
URL: https://github.com/apache/ignite-3/pull/7731#discussion_r2904766177
##########
modules/raft/src/main/java/org/apache/ignite/raft/jraft/core/NodeImpl.java:
##########
@@ -2638,6 +2638,9 @@ public Message handleAppendEntriesRequest(final
AppendEntriesRequest request, fi
done,
this.currTerm
);
+ if (request.timestamp() != null) {
+ clock.update(request.timestamp());
Review Comment:
Propagating HLC in each AppendEntry request creates additinal contention on
HLC.
For this reason it was removed earlier.
Could we instead do it in RequestVoteRequest ?
##########
modules/raft/src/integrationTest/java/org/apache/ignite/raft/server/ItNewLeaderClockTest.java:
##########
@@ -0,0 +1,199 @@
+/*
+ * 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.raft.server;
+
+import static org.apache.ignite.internal.raft.server.RaftGroupOptions.defaults;
+import static org.apache.ignite.raft.jraft.test.TestUtils.waitForCondition;
+import static
org.apache.ignite.raft.server.counter.GetValueCommand.getValueCommand;
+import static
org.apache.ignite.raft.server.counter.IncrementAndGetCommand.incrementAndGetCommand;
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertNotNull;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import org.apache.ignite.internal.hlc.HybridClock;
+import org.apache.ignite.internal.hlc.HybridClockImpl;
+import org.apache.ignite.internal.hlc.HybridTimestamp;
+import org.apache.ignite.internal.raft.Peer;
+import org.apache.ignite.internal.raft.RaftNodeId;
+import org.apache.ignite.internal.raft.server.RaftGroupOptions;
+import org.apache.ignite.internal.raft.server.RaftServer;
+import org.apache.ignite.internal.raft.server.impl.JraftServerImpl;
+import org.apache.ignite.internal.raft.service.RaftGroupService;
+import org.apache.ignite.internal.raft.util.ThreadLocalOptimizedMarshaller;
+import org.apache.ignite.internal.replicator.TestReplicationGroupId;
+import org.apache.ignite.raft.server.counter.CounterListener;
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.api.Timeout;
+
+/**
+ * Integration test for verifying that a new leader's clock doesn't step down
after leadership transfer.
+ */
+class ItNewLeaderClockTest extends JraftAbstractTest {
+ private static final TestReplicationGroupId TEST_GROUP = new
TestReplicationGroupId("leader_with_advanced_clock_step_down");
+
+ private static final HybridClock slowClock = new HybridClockImpl() {
+ @Override
+ protected long physicalTime() {
+ return 0;
+ }
+ };
+ private static final HybridClock advancedClock = new HybridClockImpl();
+
+ /**
+ * Starts a 3-node Raft cluster where two nodes have slow clock.
+ */
+ private void startClusterWithSlowClock() throws Exception {
+ // Start first server with slow clock.
+ startServer(0, raftServer -> {
+ String localNodeName =
raftServer.clusterService().topologyService().localMember().name();
+
+ Peer serverPeer = initialMembersConf.peer(localNodeName);
+
+ RaftGroupOptions groupOptions = groupOptions(raftServer);
+
+ groupOptions.setLogStorageManager(logStorageFactories.get(0));
+ groupOptions.serverDataPath(serverWorkingDirs.get(0).metaPath());
+
+ raftServer.startRaftNode(
+ new RaftNodeId(TEST_GROUP, serverPeer),
+ initialMembersConf,
+ new CounterListener(),
+ groupOptions
+ );
+ }, opts -> {
+ opts.setClock(slowClock);
+ });
+ // Start second clock with advanced clock. This node is initial leader.
+ startServer(1, raftServer -> {
+ String localNodeName =
raftServer.clusterService().topologyService().localMember().name();
+
+ Peer serverPeer = initialMembersConf.peer(localNodeName);
+
+ RaftGroupOptions groupOptions = groupOptions(raftServer);
+
+ groupOptions.setLogStorageManager(logStorageFactories.get(1));
+ groupOptions.serverDataPath(serverWorkingDirs.get(1).metaPath());
+
+ raftServer.startRaftNode(
+ new RaftNodeId(TEST_GROUP, serverPeer),
+ initialMembersConf,
+ new CounterListener(),
+ groupOptions
+ );
+ }, opts -> {
+ opts.setClock(advancedClock);
+ });
+
+ // Start third server with slow clock.
+ startServer(2, raftServer -> {
+ String localNodeName =
raftServer.clusterService().topologyService().localMember().name();
+
+ Peer serverPeer = initialMembersConf.peer(localNodeName);
+
+ RaftGroupOptions groupOptions = groupOptions(raftServer);
+
+ groupOptions.setLogStorageManager(logStorageFactories.get(2));
+ groupOptions.serverDataPath(serverWorkingDirs.get(2).metaPath());
+
+ raftServer.startRaftNode(
+ new RaftNodeId(TEST_GROUP, serverPeer),
+ initialMembersConf,
+ new CounterListener(),
+ groupOptions
+ );
+ }, opts -> {
+ opts.setClock(slowClock);
+ });
+
+ startClient(TEST_GROUP);
+ }
+
+ /**
+ * Tests that a new leader's clock doesn't step down after leadership
transfer.
+ * Verifies that:
+ * 1. Writes succeed with node 1 as leader (regular clock)
+ * 2. After stopping node 1, a new leader is elected (node 0 or 2 with
slow clock)
+ * 3. The new leader's clock is not lower than the last applied command
time
+ *
+ * @throws Exception If test fails.
+ */
+ @Test
+ @Timeout(120)
+ public void testNewLeaderClockDoesNotStepDown() throws Exception {
+ startClusterWithSlowClock();
+
+ RaftGroupService client = clients.get(0);
+
+ client.refreshLeader().get();
+ assertNotNull(client.leader(), "Initial leader should be elected");
+
+ // Force leadership to node 1 (server index 1).
+ String node1Name =
servers.get(1).clusterService().topologyService().localMember().name();
+ Peer node1Peer = initialMembersConf.peer(node1Name);
+ client.transferLeadership(node1Peer).get();
+
+ assertTrue(
+ waitForCondition(() -> {
+ client.refreshLeader().join();
+ return node1Peer.equals(client.leader());
+ }, 10_000),
+ "Leadership should transfer to node 1 with advanced clock"
+ );
+ long expectedValue = 0;
+ long currentTimestamp = System.currentTimeMillis() * 2;
+ for (int i = 0; i < 100; i++) {
+ long increment = i + 1;
+ expectedValue += increment;
+ HybridTimestamp highTime =
HybridTimestamp.hybridTimestamp(currentTimestamp--);
+ client.<Long>run(incrementAndGetCommand(increment,
highTime)).get();
+ }
+ // Verify the writes succeeded.
+ Long valueBeforeUnisolate = client.<Long>run(getValueCommand()).get();
+ assertEquals(expectedValue, valueBeforeUnisolate, "All writes should
have succeeded");
+ HybridTimestamp latestLeaderTime =
advancedClock.update(HybridTimestamp.MIN_VALUE);;
Review Comment:
```suggestion
HybridTimestamp latestLeaderTime =
advancedClock.update(HybridTimestamp.MIN_VALUE);
```
--
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]
