Skip to content

[fix][broker] Avoid load shedding and metadata writes from a former leader #26253

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

Open
void-ptr974 wants to merge 2 commits into
base: master
Choose a base branch
from
+140 −21
Open

[fix][broker] Avoid load shedding and metadata writes from a former leader #26253

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d5e269e
[fix][broker] Avoid load shedding and metadata writes from a former l…
void-ptr974 Jul 27, 2026
68869bc
[fix][broker] Stop load shedding after leadership changes
void-ptr974 Jul 29, 2026
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
57 changes: 49 additions & 8 deletions ...roker/src/main/java/org/apache/pulsar/broker/loadbalance/impl/ModularLoadManagerImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -301,11 +301,26 @@ private void handleMetadataSessionEvent(SessionEvent e) {
lastMetadataSessionEvent = e;
}

@VisibleForTesting
boolean isLeader() {
return pulsar.getLeaderElectionService() != null && pulsar.getLeaderElectionService().isLeader();
}

private LoadSheddingStrategy createLoadSheddingStrategy() {
return Reflections.createInstance(conf.getLoadBalancerLoadSheddingStrategy(), LoadSheddingStrategy.class,
Thread.currentThread().getContextClassLoader());
}

@VisibleForTesting
void setLoadSheddingStrategy(LoadSheddingStrategy loadSheddingStrategy) {
this.loadSheddingStrategy = loadSheddingStrategy;
}

@VisibleForTesting
LoadData getLoadData() {
return loadData;
}

/**
* Initialize this load manager.
*
Expand Down Expand Up @@ -635,6 +650,10 @@ public void disableBroker() throws PulsarServerException {
*/
@Override
public synchronized void doLoadShedding() {
if (!isLeader()) {
log.debug().log("Skipping load shedding because this broker is not the leader");
return;
}
if (!LoadManagerShared.isLoadSheddingEnabled(pulsar)) {
return;
}
Expand Down Expand Up @@ -684,12 +703,14 @@ public synchronized void doLoadShedding() {
return;
}

log.info().attr("class", loadSheddingStrategy.getClass().getSimpleName())
.attr("bundle", bundle).attr("sourceBroker", broker).attr("destBroker", destBroker.get())
.log("Unloading bundle from source broker to dest broker");
try {
pulsar.getAdminClient().namespaces()
.unloadNamespaceBundle(namespaceName, bundleRange, destBroker.get());
if (!isLeader()) {
return;

@Dream95 Dream95 Jul 28, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

This return is inside nested forEach lambdas, so it only skips the current foreach.This could be optimized.

@void-ptr974 void-ptr974 Jul 29, 2026

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.

Fixed in 68869bc. Replaced the nested lambdas with regular loops so leadership loss exits the load-shedding operation.

}
log.info().attr("class", loadSheddingStrategy.getClass().getSimpleName())
.attr("bundle", bundle).attr("sourceBroker", broker).attr("destBroker", destBroker.get())
.log("Unloading bundle from source broker to dest broker");
unloadNamespaceBundle(namespaceName, bundleRange, destBroker.get());
loadData.getRecentlyUnloadedBundles().put(bundle, System.currentTimeMillis());
unloadBundleCount++;
unloadBundleForBroker.set(true);
Expand All @@ -706,6 +727,12 @@ public synchronized void doLoadShedding() {
updateBundleUnloadingMetrics();
}

@VisibleForTesting
void unloadNamespaceBundle(String namespaceName, String bundleRange, String destinationBroker)
throws PulsarServerException, PulsarAdminException {
pulsar.getAdminClient().namespaces().unloadNamespaceBundle(namespaceName, bundleRange, destinationBroker);
}

/**
* As leader broker, update bundle unloading metrics.
*/
Expand Down Expand Up @@ -1204,20 +1231,34 @@ private int selectTopKBundle() {
*/
@Override
public void writeBundleDataOnZooKeeper() {
if (!isLeader()) {
log.debug().log("Skipping bundle data write because this broker is not the leader");
return;
}
updateBundleData();
if (!isLeader()) {
return;
}
// Write the bundle data to metadata store.
List<CompletableFuture<Void>> futures = new ArrayList<>();

// use synchronized to protect bundleArr.
synchronized (bundleArr) {
int updateBundleCount = selectTopKBundle();
bundleArr.stream().limit(updateBundleCount).forEach(entry -> futures.add(
pulsarResources.getLoadBalanceResources().getBundleDataResources().updateBundleData(
entry.getKey(), (BundleData) entry.getValue())));
for (Map.Entry<String, ? extends Comparable> entry : bundleArr.subList(0, updateBundleCount)) {
if (!isLeader()) {
break;
}
futures.add(pulsarResources.getLoadBalanceResources().getBundleDataResources().updateBundleData(
entry.getKey(), (BundleData) entry.getValue()));
}
}

// Write the time average broker data to metadata store.
for (Map.Entry<String, BrokerData> entry : loadData.getBrokerData().entrySet()) {
if (!isLeader()) {
break;
}
final String broker = entry.getKey();
final TimeAverageBrokerData data = entry.getValue().getTimeAverageData();
futures.add(pulsarResources.getLoadBalanceResources()
Expand Down
74 changes: 74 additions & 0 deletions ...r/src/test/java/org/apache/pulsar/broker/loadbalance/impl/ModularLoadManagerImplTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,6 +26,7 @@
import static org.mockito.Mockito.doNothing;
import static org.mockito.Mockito.spy;
import static org.mockito.Mockito.verify;
import static org.mockito.Mockito.verifyNoInteractions;
import static org.mockito.Mockito.when;
import static org.testng.Assert.assertEquals;
import static org.testng.Assert.assertFalse;
Expand All @@ -34,6 +35,7 @@
import static org.testng.Assert.fail;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.google.common.collect.BoundType;
import com.google.common.collect.ImmutableMultimap;
import com.google.common.collect.Range;
import com.google.common.collect.Sets;
import com.google.common.hash.Hashing;
Expand All @@ -56,6 +58,8 @@
import java.util.concurrent.LinkedBlockingQueue;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicReference;
import java.util.function.Supplier;
import lombok.Cleanup;
Expand All @@ -66,6 +70,7 @@
import org.apache.pulsar.broker.loadbalance.LoadBalancerTestingUtils;
import org.apache.pulsar.broker.loadbalance.LoadData;
import org.apache.pulsar.broker.loadbalance.LoadManager;
import org.apache.pulsar.broker.loadbalance.LoadSheddingStrategy;
import org.apache.pulsar.broker.loadbalance.ResourceUnit;
import org.apache.pulsar.broker.loadbalance.impl.LoadManagerShared.BrokerTopicLoadingPredicate;
import org.apache.pulsar.client.admin.Namespaces;
Expand Down Expand Up @@ -295,6 +300,75 @@ private String mockBundleName(final int i) {
return String.format("%d/%d/0x00000000_0xffffffff", i, i);
}

@Test
public void testFollowerSkipsLoadShedding() {
Awaitility.await().until(() -> pulsar1.getLeaderElectionService().isLeader()
|| pulsar2.getLeaderElectionService().isLeader());

ModularLoadManagerImpl followerLoadManager = pulsar1.getLeaderElectionService().isLeader()
? secondaryLoadManager : primaryLoadManager;

LoadSheddingStrategy loadSheddingStrategy = Mockito.mock(LoadSheddingStrategy.class);
followerLoadManager.setLoadSheddingStrategy(loadSheddingStrategy);

followerLoadManager.doLoadShedding();

verifyNoInteractions(loadSheddingStrategy);
}

@Test
public void testLoadSheddingStopsWhenLeadershipChangesBeforeUnload() throws Exception {
Awaitility.await().until(() -> primaryLoadManager.getAvailableBrokers().size() > 1);

AtomicBoolean leader = new AtomicBoolean(true);
ModularLoadManagerImpl loadManagerSpy = spy(primaryLoadManager);
doAnswer(invocation -> leader.get()).when(loadManagerSpy).isLeader();

LoadSheddingStrategy loadSheddingStrategy = Mockito.mock(LoadSheddingStrategy.class);
loadManagerSpy.setLoadSheddingStrategy(loadSheddingStrategy);
when(loadSheddingStrategy.findBundlesForUnloading(any(), any()))
.thenReturn(ImmutableMultimap.of(primaryBrokerId, mockBundleName(1)));
doAnswer(invocation -> true).when(loadManagerSpy).shouldNamespacePoliciesUnload(
Mockito.anyString(), Mockito.anyString(), Mockito.anyString());
doAnswer(invocation -> true).when(loadManagerSpy).shouldAntiAffinityNamespaceUnload(
Mockito.anyString(), Mockito.anyString(), Mockito.anyString());
doAnswer(invocation -> {
leader.set(false);
return Optional.of(secondaryBrokerId);
}).when(loadManagerSpy).selectBroker(any());
doNothing().when(loadManagerSpy).unloadNamespaceBundle(
Mockito.anyString(), Mockito.anyString(), Mockito.anyString());

loadManagerSpy.doLoadShedding();

verify(loadManagerSpy).selectBroker(any());
verify(loadManagerSpy, Mockito.never()).unloadNamespaceBundle(
Mockito.anyString(), Mockito.anyString(), Mockito.anyString());
}

@Test
public void testBundleDataWriteStopsWhenLeadershipChangesBeforeMetadataWrite() throws Exception {
String bundle = mockBundleName(99);
BundleData bundleData = new BundleData(10, 1000);
String bundleDataPath = String.format("%s/%s", BUNDLE_DATA_BASE_PATH, bundle);
MetadataCache<BundleData> metadataCache = pulsar1.getLocalMetadataStore().getMetadataCache(BundleData.class);
metadataCache.create(bundleDataPath, bundleData).join();

Awaitility.await().until(() -> primaryLoadManager.getLoadData().getBrokerData().containsKey(primaryBrokerId));
AtomicInteger leaderChecks = new AtomicInteger();
ModularLoadManagerImpl loadManagerSpy = spy(primaryLoadManager);
LoadData loadData = loadManagerSpy.getLoadData();
loadData.getBundleData().clear();
loadData.getBundleData().put(bundle, bundleData);
loadData.getBrokerData().get(primaryBrokerId).getLocalData().getLastStats()
.put(bundle, new NamespaceBundleStats());
doAnswer(invocation -> leaderChecks.getAndIncrement() < 2).when(loadManagerSpy).isLeader();

loadManagerSpy.writeBundleDataOnZooKeeper();

assertEquals(metadataCache.getWithStats(bundleDataPath).get().get().getStat().getVersion(), 0);
}

// Test disabled since it's depending on CPU usage in the machine
@Test(enabled = false)
public void testCandidateConsistency() throws Exception {
Expand Down