Skip to content

KAFKA-19500: kafka-consumer-groups.sh should fail quickly if the partition leader is unavailable #20168

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
AndrewJSchofield merged 6 commits into from
Jul 21, 2025
Merged

KAFKA-19500: kafka-consumer-groups.sh should fail quickly if the partition leader is unavailable #20168

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
157d929
finish coding
xijiu Jul 15, 2025
5084928
fix failed test
xijiu Jul 15, 2025
edeaa8a
optimize
xijiu Jul 16, 2025
3efa3bb
fix
xijiu Jul 17, 2025
bb09668
Merge branch 'trunk' of github.com:xijiu/kafka into 19500
xijiu Jul 18, 2025
dc58674
fix
xijiu Jul 18, 2025
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
32 changes: 28 additions & 4 deletions tools/src/main/java/org/apache/kafka/tools/consumer/group/ConsumerGroupCommand.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,7 @@
import org.apache.kafka.common.TopicPartition;
import org.apache.kafka.common.errors.GroupIdNotFoundException;
import org.apache.kafka.common.errors.LeaderNotAvailableException;
import org.apache.kafka.common.errors.UnknownTopicOrPartitionException;
import org.apache.kafka.common.protocol.Errors;
import org.apache.kafka.common.utils.Utils;
import org.apache.kafka.server.util.CommandLineUtils;
Expand Down Expand Up @@ -1001,8 +1002,8 @@ private Map<TopicPartition, OffsetAndMetadata> getCommittedOffsets(String groupI
}

private Map<TopicPartition, OffsetAndMetadata> prepareOffsetsToReset(String groupId, Collection<TopicPartition> partitionsToReset) {
// ensure all partitions have leader, otherwise throw a runtime exception
checkAllTopicPartitionsHaveLeader(partitionsToReset);
// ensure all partitions are valid, otherwise throw a runtime exception
checkAllTopicPartitionsValid(partitionsToReset);

if (opts.options.has(opts.resetToOffsetOpt)) {
return offsetsUtils.resetToOffset(partitionsToReset);
Expand All @@ -1028,15 +1029,38 @@ private Map<TopicPartition, OffsetAndMetadata> prepareOffsetsToReset(String grou
return null;
}

private void checkAllTopicPartitionsHaveLeader(Collection<TopicPartition> partitionsToReset) {
private void checkAllTopicPartitionsValid(Collection<TopicPartition> partitionsToReset) {
// check the partitions exist
List<TopicPartition> partitionsNotExistList = filterNotExistPartitions(partitionsToReset);
if (!partitionsNotExistList.isEmpty()) {
String partitionStr = partitionsNotExistList.stream().map(TopicPartition::toString).collect(Collectors.joining(","));
throw new UnknownTopicOrPartitionException("The partitions \"" + partitionStr + "\" does not exist");

@AndrewJSchofield AndrewJSchofield Jul 18, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

nit: Grammar "do not exist" I think.

}

// check the partitions have leader
List<TopicPartition> partitionsWithoutLeader = filterNoneLeaderPartitions(partitionsToReset);

@chia7712 chia7712 Jul 22, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

if there is a topic having three partitions, and the t-2 partition is offline, then if partitionsToReset is t-0,t-1, filterNoneLeaderPartitions will return t-2, causing the tool to fail. Is it expected?

@AndrewJSchofield AndrewJSchofield Jul 22, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

I think it's fair enough.

@chia7712 chia7712 Jul 24, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

chia7712@chia7712-ubuntu:~/project/kafka$ ./bin/kafka-consumer-groups.sh \
  --bootstrap-server 172.20.10.2:20001 \
  --reset-offsets \
  --to-earliest \
  --execute \
  --group perf-consumer-19460 \
  --topic chia:1

Error: Executing consumer group command failed due to The partitions "chia-2" have no leader
org.apache.kafka.common.errors.LeaderNotAvailableException: The partitions "chia-2" have no leader

It is indeed a bug that unrelated topic partitions could fail the tool.

@chia7712 chia7712 Jul 24, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

filterNoneLeaderPartitions needs to get fixed since it could return unrelated topic partitions.

@AndrewJSchofield AndrewJSchofield Jul 24, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

Yes, that's a very good point. Thanks for finding it.

@TaiJuWu TaiJuWu Jul 24, 2025

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

file #20235 to fix it.

if (!partitionsWithoutLeader.isEmpty()) {
String partitionStr = partitionsWithoutLeader.stream().map(TopicPartition::toString).collect(Collectors.joining(","));
// throw exception
throw new LeaderNotAvailableException("The partitions \"" + partitionStr + "\" have no leader");
}
}

private List<TopicPartition> filterNotExistPartitions(Collection<TopicPartition> topicPartitions) {

@AndrewJSchofield AndrewJSchofield Jul 18, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

nit: filterNonExistentPartitions is probably a bit better.

// collect all topics
Set<String> topics = topicPartitions.stream().map(TopicPartition::topic).collect(Collectors.toSet());
try {
List<TopicPartition> existPartitions = adminClient.describeTopics(topics).allTopicNames().get().entrySet()
.stream()
.flatMap(entry -> entry.getValue().partitions().stream()
.map(partitionInfo -> new TopicPartition(entry.getKey(), partitionInfo.partition())))
.toList();

return topicPartitions.stream().filter(element -> !existPartitions.contains(element)).toList();
} catch (Exception e) {

@AndrewJSchofield AndrewJSchofield Jul 18, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

For consistency, we tend to catch (InterruptedException | ExecutionException e) in this source file.

throw new RuntimeException(e);
}
}

String exportOffsetsToCsv(Map<String, Map<TopicPartition, OffsetAndMetadata>> assignments) {
boolean isSingleGroupQuery = opts.options.valuesOf(opts.groupOpt).size() == 1;
ObjectWriter csvWriter = isSingleGroupQuery
Expand Down
3 changes: 2 additions & 1 deletion tools/src/test/java/org/apache/kafka/tools/consumer/group/ConsumerGroupServiceTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -236,7 +236,7 @@ public void testAdminRequestsForResetOffsets() {
when(admin.describeTopics(ArgumentMatchers.eq(topicsWithoutPartitionsSpecified), any()))
.thenReturn(describeTopicsResult(topicsWithoutPartitionsSpecified));
when(admin.describeTopics(anySet()))
.thenReturn(describeTopicsResult(topicsWithoutPartitionsSpecified));
.thenReturn(describeTopicsResult(TOPICS));
when(admin.listOffsets(offsetsArgMatcher(), any()))
.thenReturn(listOffsetsResult());

Expand Down Expand Up @@ -324,6 +324,7 @@ private DescribeTopicsResult describeTopicsResult(Collection<String> topics) {
.collect(Collectors.toList());
topicDescriptions.put(topic, new TopicDescription(topic, false, partitions));
});
System.out.println("cccc: " + topicDescriptions);
return AdminClientTestUtils.describeTopicsResult(topicDescriptions);
}

Expand Down
17 changes: 17 additions & 0 deletions tools/src/test/java/org/apache/kafka/tools/consumer/group/ResetConsumerGroupOffsetTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,7 @@
import org.apache.kafka.common.GroupState;
import org.apache.kafka.common.TopicPartition;
import org.apache.kafka.common.errors.LeaderNotAvailableException;
import org.apache.kafka.common.errors.UnknownTopicOrPartitionException;
import org.apache.kafka.common.serialization.ByteArraySerializer;
import org.apache.kafka.common.serialization.StringDeserializer;
import org.apache.kafka.common.test.ClusterInstance;
Expand Down Expand Up @@ -680,6 +681,22 @@ public void testResetOffsetsWithPartitionNoneLeader(ClusterInstance cluster) thr
}
}

@ClusterTest
public void testResetOffsetsWithPartitionNotExist(ClusterInstance cluster) throws Exception {
String group = generateRandomGroupId();
String topic = generateRandomTopic();
String[] args = buildArgsForGroup(cluster, group, "--topic", topic + ":2,3",
"--to-earliest", "--execute");

try (Admin admin = cluster.admin();
ConsumerGroupCommand.ConsumerGroupService service = getConsumerGroupService(args)) {

admin.createTopics(singleton(new NewTopic(topic, 1, (short) 1))).all().get();
produceConsumeAndShutdown(cluster, topic, group, 2, GroupProtocol.CLASSIC);
assertThrows(UnknownTopicOrPartitionException.class, () -> resetOffsets(service));
}
}

private String generateRandomTopic() {
return TOPIC_PREFIX + TestUtils.randomString(10);
}
Expand Down
Loading