Apache Kafka 2.1.1 发布了，这是一个 bugfix 版本。
更新内容如下：
Improvement
[KAFKA-6928] - StreamsPartitionAssignor is double retrying within InternalTopicManager
[KAFKA-7389] - Upgrade spotBugs for Java 11 support
[KAFKA-7584] - StreamsConfig throws ClassCastException if max.in.flight.request.per.connect is specified as String
[KAFKA-7610] - Detect consumer failures in initial JoinGroup
[KAFKA-7671] - A KStream/GlobalKTable join shouldn't reset the repartition flag
[KAFKA-7734] - Metrics tags should use LinkedHashMap to guarantee ordering
[KAFKA-7837] - maybeShrinkIsr may not reflect OfflinePartitions immediately
Bug
[KAFKA-7897] - Invalid use of epoch cache with old message format versions
[KAFKA-3832] - Kafka Connect's JSON Converter never outputs a null value
[KAFKA-5117] - Kafka Connect REST endpoints reveal Password typed values
[KAFKA-6388] - Error while trying to roll a segment that already exists
[KAFKA-6833] - KafkaProducer throws "Invalid partition given with record" exception
[KAFKA-7253] - The connector type responded by worker is always null when creating connector
[KAFKA-7443] - OffsetOutOfRangeException in restoring state store from changelog topic when start offset of local checkpoint is smaller than that of changelog topic
[KAFKA-7518] - FutureRecordMetadata.get deadline calculation from timeout is not using timeunit
[KAFKA-7532] - Missing stringification in error message "Removed ArrayBuffer() from list of shutting down brokers. (kafka.controller.KafkaController)"
[KAFKA-7536] - TopologyTestDriver cannot pre-populate KTable or GlobalKTable
[KAFKA-7549] - Old ProduceRequest with zstd compression does not return error to client
[KAFKA-7576] - Dynamic update of replica fetcher threads may fail to start/close fetchers
[KAFKA-7607] - NetworkClientUtils.sendAndReceive can take a long time to return during shutdown
[KAFKA-7620] - ConfigProvider is broken for KafkaConnect when TTL is not null
[KAFKA-7655] - Metadata spamming requests from Kafka Streams under some circumstances, potential DOS
[KAFKA-7660] - Stream Metrics - Memory Analysis
[KAFKA-7678] - Failed to close producer due to java.lang.NullPointerException
[KAFKA-7692] - updateFirstUnstableOffset NPE due to sequenceId overflow in ProducerStateManager.append
[KAFKA-7693] - "IllegalArgumentException: Invalid negative sequence number used" in Kafka Client
[KAFKA-7697] - Possible deadlock in kafka.cluster.Partition
[KAFKA-7702] - Prefixed ACLs don't work with single character prefix
[KAFKA-7704] - kafka.server.ReplicaFetechManager.MaxLag.Replica metric is reported incorrectly
[KAFKA-7705] - Update javadoc for the values of delivery.timeout.ms or linger.ms
[KAFKA-7709] - ConcurrentModificationException occurs when iterating through multiple partitions in Sender.getExpiredInflightBatches
[KAFKA-7741] - Bad dependency via SBT
[KAFKA-7755] - Kubernetes - Kafka clients are resolving DNS entries only one time
[KAFKA-7759] - Disable WADL output in Connect REST API
[KAFKA-7768] - Java doc link 404
[KAFKA-7786] - Fast update of leader epoch may stall partition fetching due to FENCED_LEADER_EPOCH
[KAFKA-7840] - Documentation for cleanup.policy is out of date
[KAFKA-7873] - KafkaBasedLog's consumer should always seek to beginning when starting
[KAFKA-7890] - Invalidate ClusterConnectionState cache for a broker if the hostname of the broker changes.
[KAFKA-7902] - SASL/OAUTHBEARER can become unable to connect: javax.security.sasl.SaslException: Unable to find OAuth Bearer token in Subject's private credentials (size=2)
Task
[KAFKA-7503] - Integration Test Framework for Connect
Sub-task
[KAFKA-5503] - Idempotent producer ignores shutdown while fetching ProducerId
Kafka是一种高吞吐量的分布式发布订阅消息系统，她有如下特性：
通过O(1)的磁盘数据结构提供消息的持久化，这种结构对于即使数以TB的消息存储也能够保持长时间的稳定性能。
高吞吐量：即使是非常普通的硬件kafka也可以支持每秒数十万的消息。
支持通过kafka服务器和消费机集群来分区消息。
支持Hadoop并行数据加载。
下载地址：https://kafka.apache.org/downloads#2.1.1