Skip to content

fix(zerozone2): fix reset bug cause skip #2853

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

Merged
merged 1 commit into from
Sep 16, 2025
Merged

fix(zerozone2): fix reset bug cause skip #2853

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 5 additions & 5 deletions core/src/main/java/kafka/automq/zerozone/SnapshotReadPartitionsManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -305,7 +305,7 @@ void unsafeRun() {
this.run0();
} catch (Throwable e) {
LOGGER.error("[SNAPSHOT_SUBSCRIBE_ERROR]", e);
reset();
reset("SUBSCRIBE_ERROR: " + e.getMessage());
scheduler.schedule(this::run, 1, TimeUnit.SECONDS);
}
}
Expand All @@ -321,8 +321,8 @@ private void run0() {
applySnapshot();
}

void reset() {
LOGGER.info("[SNAPSHOT_READ_SUBSCRIBER_RESET],node={}", node);
void reset(String reason) {
LOGGER.info("[SNAPSHOT_READ_SUBSCRIBER_RESET],node={},reason={}", node, reason);
partitions.forEach(SnapshotReadPartitionsManager.this::removePartition);
partitions.clear();
waitingMetadataReadyQueue.clear();
Expand All @@ -348,7 +348,7 @@ void applySnapshot() {
case ADD: {
Optional<Partition> partition = addPartition(topicIdPartition, snapshotWithOperation.snapshot);
if (partition.isEmpty()) {
reset();
reset(String.format("Cannot find partition %s", topicIdPartition));
return;
}
partition.ifPresent(p -> partitions.put(topicIdPartition, p));
Expand All @@ -360,7 +360,7 @@ void applySnapshot() {
if (partition != null) {
partition.snapshot(snapshotWithOperation.snapshot);
} else {
LOGGER.warn("[SNAPSHOT_READ_PATCH],[SKIP],{}", snapshotWithOperation);
LOGGER.error("[SNAPSHOT_READ_PATCH],[SKIP],{}", snapshotWithOperation);
}
snapshotWithOperations.poll();
break;
Expand Down
29 changes: 23 additions & 6 deletions core/src/main/java/kafka/automq/zerozone/SubscriberRequester.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -59,6 +59,7 @@
private int sessionId;
private int sessionEpoch;
boolean requestCommit = false;
boolean requestReset = false;

private final SnapshotReadPartitionsManager.Subscriber subscriber;
private final Node node;
Expand All @@ -81,8 +82,7 @@ public SubscriberRequester(SnapshotReadPartitionsManager.Subscriber subscriber,
}

public void reset() {
sessionId = 0;
sessionEpoch = 0;
requestReset = true;
}

public void close() {
Expand All @@ -97,6 +97,7 @@ private void request0() {
if (closed) {
return;
}
tryReset0();
lastRequestTime = time.milliseconds();
AutomqGetPartitionSnapshotRequestData data = new AutomqGetPartitionSnapshotRequestData().setSessionId(sessionId).setSessionEpoch(sessionEpoch).setVersion((short) 1);
if (version.isZeroZoneV2Supported()) {
Expand Down Expand Up @@ -132,6 +133,10 @@ private void handleResponse(ClientResponse clientResponse) {
if (closed) {
return;
}
if (tryReset0()) {
// If it needs to reset, then drop the response.
return;
}
if (!clientResponse.hasResponse()) {
if (clientResponse.wasDisconnected() || clientResponse.wasTimedOut()) {
LOGGER.warn("[GET_SNAPSHOTS],[REQUEST_FAIL],response={}", clientResponse);
Expand All @@ -149,16 +154,17 @@ private void handleResponse(ClientResponse clientResponse) {
LOGGER.error("[GET_SNAPSHOTS],[ERROR],response={}", resp);
return;
}
if (resp.sessionId() != sessionId) {
if (sessionId != 0 && resp.sessionId() != sessionId) {
// switch to a new session
subscriber.reset();
subscriber.reset(String.format("switch sessionId from %s to %s", sessionId, resp.sessionId()));
}
SnapshotReadPartitionsManager.OperationBatch batch = new SnapshotReadPartitionsManager.OperationBatch();
resp.topics().forEach(topic -> topic.partitions().forEach(partition -> {
String topicName = topicNameGetter.apply(topic.topicId());
if (topicName == null) {
subscriber.reset();
throw new RuntimeException(String.format("Cannot find topic uuid=%s, the kraft metadata replay delay or the topic is deleted.", topic.topicId()));
String reason = String.format("Cannot find topic uuid=%s, the kraft metadata replay delay or the topic is deleted.", topic.topicId());
subscriber.reset(reason);
throw new RuntimeException(reason);
}
batch.operations.add(convert(new TopicIdPartition(topic.topicId(), partition.partitionIndex(), topicName), partition));
}));
Expand All @@ -174,6 +180,17 @@ private void handleResponse(ClientResponse clientResponse) {
sessionEpoch = resp.sessionEpoch();
}

private boolean tryReset0() {
if (requestReset) {
sessionId = 0;
sessionEpoch = 0;
requestReset = false;
return true;
} else {
return false;
}
}

static SnapshotReadPartitionsManager.SnapshotWithOperation convert(TopicIdPartition topicIdPartition,
AutomqGetPartitionSnapshotResponseData.PartitionSnapshot src) {
PartitionSnapshot.Builder snapshot = PartitionSnapshot.builder();
Expand Down
Loading