这里有很多很棒的答案，但在其中，我没有找到一个关于docker的答案。我花了一些时间来弄清楚在这种情况下使用代理容器是错误的(显然!!)

## this is wrong!
docker exec broker1 kafka-topics --zookeeper localhost:2181 --alter --topic mytopic --config retention.ms=1000

Exception in thread "main" kafka.zookeeper.ZooKeeperClientTimeoutException: Timed out waiting for connection while in state: CONNECTING
        at kafka.zookeeper.ZooKeeperClient.$anonfun$waitUntilConnected$3(ZooKeeperClient.scala:258)
        at scala.runtime.java8.JFunction0$mcV$sp.apply(JFunction0$mcV$sp.java:23)
        at kafka.utils.CoreUtils$.inLock(CoreUtils.scala:253)
        at kafka.zookeeper.ZooKeeperClient.waitUntilConnected(ZooKeeperClient.scala:254)
        at kafka.zookeeper.ZooKeeperClient.<init>(ZooKeeperClient.scala:112)
        at kafka.zk.KafkaZkClient$.apply(KafkaZkClient.scala:1826)
        at kafka.admin.TopicCommand$ZookeeperTopicService$.apply(TopicCommand.scala:280)
        at kafka.admin.TopicCommand$.main(TopicCommand.scala:53)
        at kafka.admin.TopicCommand.main(TopicCommand.scala)

我应该使用zookeeper:2181而不是——zookeeper localhost:2181作为我的撰写文件

## this might be an option, but as per comment below not all zookeeper images can have this script included
docker exec zookeper1 kafka-topics --zookeeper localhost:2181 --alter --topic mytopic --config retention.ms=1000

正确的命令应该是

docker exec broker1 kafka-configs --zookeeper zookeeper:2181 --alter --entity-type topics --entity-name dev_gdn_urls --add-config retention.ms=12800000

希望能节省一些人的时间。

另外，请注意消息不会立即删除，而是在关闭日志段时删除。

更新:这个答案与Kafka 0.6相关。对于Kafka 0.8和以后参见@Patrick的回答。

是的，停止kafka，手动删除相应子目录下的所有文件(在kafka数据目录下很容易找到)。kafka重启后，主题将为空。

如果您希望在Java应用程序中以编程方式完成此操作，可以使用AdminClient的API deleterrecords。使用AdminClient可以删除分区和偏移量级别上的记录。

根据JavaDocs, 0.11.0.0或更高版本的代理支持此操作。

这里有一个简单的例子:

String brokers = "localhost:9092";
String topicName = "test";
TopicPartition topicPartition = new TopicPartition(topicName, 0);
RecordsToDelete recordsToDelete = RecordsToDelete.beforeOffset(5L);

Map<TopicPartition, RecordsToDelete> topicPartitionRecordToDelete = new HashMap<>();
topicPartitionRecordToDelete.put(topicPartition, recordsToDelete);

// Create AdminClient
final Properties properties = new Properties();
properties.put(AdminClientConfig.BOOTSTRAP_SERVERS_CONFIG, brokers);
AdminClient adminClient = AdminClient.create(properties);

try {
  adminClient.deleteRecords(topicPartitionRecordToDelete).all().get();
} catch (InterruptedException e) {
  e.printStackTrace();
} catch (ExecutionException e) {
  e.printStackTrace();
} finally {
  adminClient.close();
}

您必须在配置中启用此功能

echo "delete.topic.enable=true" >> /opt/kafka/config/server.properties 
sudo systemctl stop kafka 
sudo systemctl start kafka 

清除主题

/opt/kafka/bin/kafka-topics.sh --bootstrap-server localhost:9092 --delete --topic flows

创建主题

# /opt/kafka/bin/kafka-topics.sh --create --bootstrap-server localhost:2181 --replication-factor 1 --partitions 1 --topic Test

阅读主题

# /opt/kafka/bin/kafka-console-consumer.sh  localhost:9092 --topic flows --from-beginning

最简单的方法是将各个日志文件的日期设置为比保留期更早的日期。然后经纪人会在几秒钟内为你清理并移除它们。这有几个优点:

不需要关闭代理，这是一个运行时操作。 避免出现无效偏移异常的可能性(下文将详细介绍)。

In my experience with Kafka 0.7.x, removing the log files and restarting the broker could lead to invalid offset exceptions for certain consumers. This would happen because the broker restarts the offsets at zero (in the absence of any existing log files), and a consumer that was previously consuming from the topic would reconnect to request a specific [once valid] offset. If this offset happens to fall outside the bounds of the new topic logs, then no harm and the consumer resumes at either the beginning or the end. But, if the offset falls within the bounds of the new topic logs, the broker attempts to fetch the message set but fails because the offset doesn't align to an actual message.

还可以通过清除zookeeper中针对该主题的消费者偏移量来缓解这一问题。但如果你不需要一个处女主题，只是想删除现有的内容，那么简单地“触摸”一些主题日志要比停止代理、删除主题日志和清除某些zookeeper节点容易得多，也更可靠。

我使用的是Kafka 2.13工具。现在——zookeeper是kafka-topics.sh的不可识别选项。删除主题。

bin/kafka-topics.sh --bootstrap-server [kafka broker]:9092 --delete --topic [topic name]

只是要考虑到，如果你在删除的主题中有很多数据，再次创建相同的主题可能需要一段时间。当你尝试创建相同的主题时，你可能会得到这样的错误:

Topic: Topic “[主题名]”被标记为删除。