Skip to content

HDDS-14070. set statemachine ready on election performed for follower #9671

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
sumitagrawl wants to merge 2 commits into
base: master
Choose a base branch
from
+25 −4
Open

HDDS-14070. set statemachine ready on election performed for follower #9671

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7c1ece2
HDDS-14070. set statemachine ready on election performed for follower
sumitagrawl Jan 26, 2026
42dee1d
test case added
sumitagrawl Jan 26, 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
14 changes: 10 additions & 4 deletions hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/ha/SCMStateMachine.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -285,6 +285,15 @@ public void notifyLeaderChanged(RaftGroupMemberId groupMemberId,
currentLeaderTerm.set(scm.getScmHAManager().getRatisServer().getDivision()
.getInfo().getCurrentTerm());

if (!refreshedAfterLeaderReady.get()) {
// refresh and validate safe mode rules if it can exit safe mode
// if being leader, all previous term transactions have been applied
// if other states, just refresh safe mode rules, and transaction keeps flushing from leader
// and does not depend on pending transactions.
refreshedAfterLeaderReady.set(true);
Comment on lines +288 to +293
Copy link
Contributor

@szetszwo szetszwo Feb 1, 2026

Choose a reason for hiding this comment

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

This is not atomic. Use

    if (refreshedAfterLeaderReady.compareAndSet(false, true)) {

scm.getScmSafeModeManager().refreshAndValidate();
}

if (!groupMemberId.getPeerId().equals(newLeaderId)) {
LOG.info("leader changed, yet current SCM is still follower.");
return;
Expand Down Expand Up @@ -355,12 +364,9 @@ public void notifyTermIndexUpdated(long term, long index) {
}

if (currentLeaderTerm.get() == term) {
// Means all transactions before this term have been applied.
// This means after a restart, all pending transactions have been applied.
// Perform
// 1. Refresh Safemode rules state.
// 2. Start DN Rpc server.
if (!refreshedAfterLeaderReady.get()) {
Comment on lines 366 to 368
Copy link
Contributor

@szetszwo szetszwo Feb 1, 2026

Choose a reason for hiding this comment

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

Similarly, it needs compareAndSet.

BTW, why refreshedAfterLeaderReady won't be set to false after set to true?

// Refresh Safemode rules state if not already done.
refreshedAfterLeaderReady.set(true);
scm.getScmSafeModeManager().refreshAndValidate();
}
Expand Down
15 changes: 15 additions & 0 deletions .../server-scm/src/test/java/org/apache/hadoop/hdds/scm/safemode/TestSCMSafeModeManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,8 +54,11 @@
import org.apache.hadoop.hdds.scm.container.replication.ContainerReplicaPendingOps;
import org.apache.hadoop.hdds.scm.events.SCMEvents;
import org.apache.hadoop.hdds.scm.ha.SCMContext;
import org.apache.hadoop.hdds.scm.ha.SCMHAManager;
import org.apache.hadoop.hdds.scm.ha.SCMHAManagerStub;
import org.apache.hadoop.hdds.scm.ha.SCMRatisServer;
import org.apache.hadoop.hdds.scm.ha.SCMServiceManager;
import org.apache.hadoop.hdds.scm.ha.SCMStateMachine;
import org.apache.hadoop.hdds.scm.metadata.SCMMetadataStore;
import org.apache.hadoop.hdds.scm.metadata.SCMMetadataStoreImpl;
import org.apache.hadoop.hdds.scm.node.NodeManager;
Expand All @@ -67,6 +70,7 @@
import org.apache.hadoop.hdds.scm.pipeline.PipelineProvider;
import org.apache.hadoop.hdds.scm.server.SCMDatanodeHeartbeatDispatcher;
import org.apache.hadoop.hdds.scm.server.SCMDatanodeProtocolServer;
import org.apache.hadoop.hdds.scm.server.StorageContainerManager;
import org.apache.hadoop.hdds.server.events.EventQueue;
import org.apache.ozone.test.GenericTestUtils;
import org.junit.jupiter.api.AfterEach;
Expand Down Expand Up @@ -177,6 +181,16 @@ public void testSafeModeExitRule() throws Exception {
}
ContainerManager containerManager = mock(ContainerManager.class);
when(containerManager.getContainers(ReplicationType.RATIS)).thenReturn(containers);

StorageContainerManager mockScmManager = mock(StorageContainerManager.class);
SCMHAManager mockScmhaManager = mock(SCMHAManager.class);
when(mockScmManager.getScmHAManager()).thenReturn(mockScmhaManager);
SCMRatisServer mockScmRatisServer = mock(SCMRatisServer.class);
when(mockScmhaManager.getRatisServer()).thenReturn(mockScmRatisServer);
SCMStateMachine mockScmStateMachine = mock(SCMStateMachine.class);
when(mockScmRatisServer.getSCMStateMachine()).thenReturn(mockScmStateMachine);
when((mockScmStateMachine.isRefreshedAfterLeaderReady())).thenReturn(true);
scmContext = new SCMContext.Builder().setSCM(mockScmManager).build();
Comment on lines +185 to +193
Copy link
Contributor

@szetszwo szetszwo Feb 1, 2026

Choose a reason for hiding this comment

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

Please don't use mock. We need to a real cluster test which test SCM changing leader multiple times.

scmSafeModeManager = new SCMSafeModeManager(config, null, null, containerManager,
serviceManager, queue, scmContext);
scmSafeModeManager.start();
Expand Down Expand Up @@ -207,6 +221,7 @@ public void testSafeModeExitRule() throws Exception {
testContainerThreshold(containers.subList(75, 100), 1.0);
assertEquals(100, scmSafeModeManager.getSafeModeMetrics()
.getCurrentContainersWithOneReplicaReportedCount().value());
scmSafeModeManager.validateSafeModeExitRules(StateMachineReadyRule.class.getSimpleName());

GenericTestUtils.waitFor(() -> !scmSafeModeManager.getInSafeMode(),
100, 1000 * 5);
Expand Down