一、副本基本信息
(1）Kafka 副本作用：提高数据可靠性。
（2）Kafka 默认副本 1 个，生产环境一般配置为 2 个，保证数据可靠性；太多副本会
增加磁盘存储空间，增加网络上数据传输，降低效率。
（3）Kafka 中副本分为：Leader 和 Follower。Kafka 生产者只会把数据发往 Leader，
然后 Follower 找 Leader 进行同步数据。
（4）Kafka 分区中的所有副本统称为 AR（Assigned Repllicas）。AR = ISR + OSR
ISR，表示和 Leader 保持同步的 Follower 集合。如果 Follower 长时间未向 Leader 发送通信请求或同步数据，则该 Follower 将被踢出 ISR。该时间阈值由 replica.lag.time.max.ms参数设定，默认 30s。Leader 发生故障之后，就会从 ISR 中选举新的 Leader。
OSR，表示 Follower 与 Leader 副本同步时，延迟过多的副本。
二、Leader 选举流程
1、简介
Kafka 集群中有一个 broker 的 Controller 会被选举为 Controller Leader，负责管理集群
broker 的上下线，所有 topic 的分区副本分配和== Leader 选举==等工作。
Controller 的信息同步工作是依赖于 Zookeeper 的。
（1）创建一个新的 topic，4 个分区，4 个副本
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --create --topic atguigu1 --partitions 4 --replication-factor 4 Created topic atguigu1. [dhapp@conch01 kafka_2.12-3.0.0]$（2）查看 Leader 分布情况
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 2 Replicas: 2,1,0,3 Isr: 2,1,0,3 Topic: atguigu1 Partition: 1 Leader: 3 Replicas: 3,0,2,1 Isr: 3,0,2,1 Topic: atguigu1 Partition: 2 Leader: 1 Replicas: 1,2,3,0 Isr: 1,2,3,0 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,3,1,2 [dhapp@conch01 kafka_2.12-3.0.0]$由上可知Isr中包含了四个节点，都是存活的
（3）停止掉 conch04 的 kafka 进程，并查看 Leader 分区情况
[dhapp@conch04 kafka_2.12-3.0.0]$ bin/kafka-server-stop.sh [dhapp@conch04 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 2 Replicas: 2,1,0,3 Isr: 2,1,0 Topic: atguigu1 Partition: 1 Leader: 0 Replicas: 3,0,2,1 Isr: 0,2,1 Topic: atguigu1 Partition: 2 Leader: 1 Replicas: 1,2,3,0 Isr: 1,2,0 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,1,2 [dhapp@conch04 kafka_2.12-3.0.0]$由上可知conch04停掉后,原本的Leader为3的也切换了0，Isr中只有2,1,0
（4）停止掉 conch03的 kafka 进程，并查看 Leader 分区情况
[dhapp@conch03 kafka_2.12-3.0.0]$ bin/kafka-server-stop.sh [dhapp@conch03 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 1 Replicas: 2,1,0,3 Isr: 1,0 Topic: atguigu1 Partition: 1 Leader: 0 Replicas: 3,0,2,1 Isr: 0,1 Topic: atguigu1 Partition: 2 Leader: 1 Replicas: 1,2,3,0 Isr: 1,0 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,1 [dhapp@conch03 kafka_2.12-3.0.0]$（5）启动 conch04的 kafka 进程，并查看 Leader 分区情况
[dhapp@conch04 kafka_2.12-3.0.0]$ bin/kafka-server-start.sh -daemon config/server.properties [dhapp@conch04 kafka_2.12-3.0.0]$ jps 3844 Jps 3813 Kafka [dhapp@conch04 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 1 Replicas: 2,1,0,3 Isr: 1,0,3 Topic: atguigu1 Partition: 1 Leader: 0 Replicas: 3,0,2,1 Isr: 0,1,3 Topic: atguigu1 Partition: 2 Leader: 1 Replicas: 1,2,3,0 Isr: 1,0,3 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,1,3 [dhapp@conch04 kafka_2.12-3.0.0]$（6）启动 conch03 的 kafka 进程，并查看 Leader 分区情况
[dhapp@conch03 kafka_2.12-3.0.0]$ bin/kafka-server-start.sh -daemon config/server.properties [dhapp@conch03 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 1 Replicas: 2,1,0,3 Isr: 1,0,3,2 Topic: atguigu1 Partition: 1 Leader: 0 Replicas: 3,0,2,1 Isr: 0,1,3,2 Topic: atguigu1 Partition: 2 Leader: 1 Replicas: 1,2,3,0 Isr: 1,0,3,2 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,1,3,2 [dhapp@conch03 kafka_2.12-3.0.0]$（7）停止掉 conch02 的 kafka 进程，并查看 Leader 分区情况
[dhapp@conch02 kafka_2.12-3.0.0]$ bin/kafka-server-stop.sh [dhapp@conch02 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic atguigu1 Topic: atguigu1 TopicId: kYmWU1iuRoGLN-fqKab99A PartitionCount: 4 ReplicationFactor: 4 Configs: segment.bytes=1073741824 Topic: atguigu1 Partition: 0 Leader: 2 Replicas: 2,1,0,3 Isr: 0,3,2 Topic: atguigu1 Partition: 1 Leader: 3 Replicas: 3,0,2,1 Isr: 0,3,2 Topic: atguigu1 Partition: 2 Leader: 2 Replicas: 1,2,3,0 Isr: 0,3,2 Topic: atguigu1 Partition: 3 Leader: 0 Replicas: 0,3,1,2 Isr: 0,3,2 [dhapp@conch02 kafka_2.12-3.0.0]$
三、Leader 和 Follower 故障处理细节
四、分区副本分配
如果 kafka 服务器只有 4 个节点，那么设置 kafka 的分区数大于服务器台数，在 kafka底层如何分配存储副本呢？
1）创建 16 分区，3 个副本
（1）创建一个新的 topic，名称为 second。[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --create --partitions 16 --replication-factor 3 --topic second Created topic second. [dhapp@conch01 kafka_2.12-3.0.0]$（2）查看分区和副本情况。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic second Topic: second TopicId: WSdQ_UR0RnOdfMrDollB3Q PartitionCount: 16 ReplicationFactor: 3 Configs: segment.bytes=1073741824 Topic: second Partition: 0 Leader: 0 Replicas: 0,3,1 Isr: 0,3,1 Topic: second Partition: 1 Leader: 2 Replicas: 2,1,0 Isr: 2,1,0 Topic: second Partition: 2 Leader: 3 Replicas: 3,0,2 Isr: 3,0,2 Topic: second Partition: 3 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3 Topic: second Partition: 4 Leader: 0 Replicas: 0,1,2 Isr: 0,1,2 Topic: second Partition: 5 Leader: 2 Replicas: 2,0,3 Isr: 2,0,3 Topic: second Partition: 6 Leader: 3 Replicas: 3,2,1 Isr: 3,2,1 Topic: second Partition: 7 Leader: 1 Replicas: 1,3,0 Isr: 1,3,0 Topic: second Partition: 8 Leader: 0 Replicas: 0,2,3 Isr: 0,2,3 Topic: second Partition: 9 Leader: 2 Replicas: 2,3,1 Isr: 2,3,1 Topic: second Partition: 10 Leader: 3 Replicas: 3,1,0 Isr: 3,1,0 Topic: second Partition: 11 Leader: 1 Replicas: 1,0,2 Isr: 1,0,2 Topic: second Partition: 12 Leader: 0 Replicas: 0,3,1 Isr: 0,3,1 Topic: second Partition: 13 Leader: 2 Replicas: 2,1,0 Isr: 2,1,0 Topic: second Partition: 14 Leader: 3 Replicas: 3,0,2 Isr: 3,0,2 Topic: second Partition: 15 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3 [dhapp@conch01 kafka_2.12-3.0.0]$分区副本分配
五、手动调整分区副本存储
手动调整分区副本存储的步骤如下：
（1）创建一个新的 topic，名称为 three。[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --create --partitions 4 --replication-factor 2 --topic three Created topic three. [dhapp@conch01 kafka_2.12-3.0.0]$（2）查看分区副本存储情况。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic three Topic: three TopicId: E95kRGLIS3C4dFCA5XcgdA PartitionCount: 4 ReplicationFactor: 2 Configs: segment.bytes=1073741824 Topic: three Partition: 0 Leader: 2 Replicas: 2,1 Isr: 2,1 Topic: three Partition: 1 Leader: 3 Replicas: 3,0 Isr: 3,0 Topic: three Partition: 2 Leader: 1 Replicas: 1,2 Isr: 1,2 Topic: three Partition: 3 Leader: 0 Replicas: 0,3 Isr: 0,3 [dhapp@conch01 kafka_2.12-3.0.0]$（3）创建副本存储计划（所有副本都指定存储在 broker0、broker1 中）。
[dhapp@conch01 kafka_2.12-3.0.0]$ vim increase-replication-factor.json [dhapp@conch01 kafka_2.12-3.0.0]$ cat increase-replication-factor.json { "version":1, "partitions":[ {"topic":"three","partition":0,"replicas":[0,1]}, {"topic":"three","partition":1,"replicas":[0,1]}, {"topic":"three","partition":2,"replicas":[1,0]}, {"topic":"three","partition":3,"replicas":[1,0]}] } [dhapp@conch01 kafka_2.12-3.0.0]$（4）执行副本存储计划。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-reassign-partitions.sh --bootstrap-server conch01:9092 --reassignment-json-file increase-replication-factor.json --execute Current partition replica assignment {"version":1,"partitions":[{"topic":"three","partition":0,"replicas":[2,1],"log_dirs":["any","any"]},{"topic":"three","partition":1,"replicas":[3,0],"log_dirs":["any","any"]},{"topic":"three","partition":2,"replicas":[1,2],"log_dirs":["any","any"]},{"topic":"three","partition":3,"replicas":[0,3],"log_dirs":["any","any"]}]} Save this to use as the --reassignment-json-file option during rollback Successfully started partition reassignments for three-0,three-1,three-2,three-3 [dhapp@conch01 kafka_2.12-3.0.0]$（5）验证副本存储计划。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-reassign-partitions.sh --bootstrap-server conch01:9092 --reassignment-json-file increase-replication-factor.json --verify Status of partition reassignment: Reassignment of partition three-0 is complete. Reassignment of partition three-1 is complete. Reassignment of partition three-2 is complete. Reassignment of partition three-3 is complete. Clearing broker-level throttles on brokers 0,1,2,3 Clearing topic-level throttles on topic three [dhapp@conch01 kafka_2.12-3.0.0]$（6）查看分区副本存储情况。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic three Topic: three TopicId: E95kRGLIS3C4dFCA5XcgdA PartitionCount: 4 ReplicationFactor: 2 Configs: segment.bytes=1073741824 Topic: three Partition: 0 Leader: 0 Replicas: 0,1 Isr: 1,0 Topic: three Partition: 1 Leader: 0 Replicas: 0,1 Isr: 0,1 Topic: three Partition: 2 Leader: 1 Replicas: 1,0 Isr: 1,0 Topic: three Partition: 3 Leader: 1 Replicas: 1,0 Isr: 0,1 [dhapp@conch01 kafka_2.12-3.0.0]$由上可以看到所有的副本都存储在broker0和broker1服务上了
六、Leader Partition 负载平衡
- auto.leader.rebalance.enable默认是 true。 自动 Leader Partition 平衡。生产环境中，leader 重选举的代价比较大，可能会带来性能影响，建议设置为 false 关闭。
- leader.imbalance.per.broker.percentage 默认是 10%。每个 broker 允许的不平衡的 leader的比率。如果每个 broker 超过了这个值，控制器会触发 leader 的平衡。
- leader.imbalance.check.interval.seconds默认值 300 秒。检查 leader 负载是否平衡的间隔时间。
七、增加副本因子
在生产环境当中，由于某个主题的重要等级需要提升，我们考虑增加副本。副本数的增加需要先制定计划，然后根据计划执行。
不能通过命令行的方法添加副本。
1）创建 topic[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --create --partitions 3 --replication-factor 1 --topic four Created topic four. [dhapp@conch01 kafka_2.12-3.0.0]$2）手动增加副本存储
（1）创建副本存储计划（所有副本都指定存储在 broker0、broker1、broker2 中）。[dhapp@conch01 kafka_2.12-3.0.0]$ vim increase-replication-factor.json [dhapp@conch01 kafka_2.12-3.0.0]$ cat increase-replication-factor.json {"version":1,"partitions":[ {"topic":"four","partition":0,"replicas":[0,1,2]}, {"topic":"four","partition":1,"replicas":[0,1,2]}, {"topic":"four","partition":2,"replicas":[0,1,2]}] } [dhapp@conch01 kafka_2.12-3.0.0]$（2）执行副本存储计划。
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-reassign-partitions.sh --bootstrap-server conch01:9092 --reassignment-json-file increase-replication-factor.json --execute Current partition replica assignment {"version":1,"partitions":[{"topic":"four","partition":0,"replicas":[3],"log_dirs":["any"]},{"topic":"four","partition":1,"replicas":[1],"log_dirs":["any"]},{"topic":"four","partition":2,"replicas":[0],"log_dirs":["any"]}]} Save this to use as the --reassignment-json-file option during rollback Successfully started partition reassignments for four-0,four-1,four-2 [dhapp@conch01 kafka_2.12-3.0.0]$（3）查看详情
[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-topics.sh --bootstrap-server conch01:9092 --describe --topic four Topic: four TopicId: WG70RfSsRDipj6VpEXgeVw PartitionCount: 3 ReplicationFactor: 3 Configs: segment.bytes=1073741824 Topic: four Partition: 0 Leader: 0 Replicas: 0,1,2 Isr: 2,0,1 Topic: four Partition: 1 Leader: 1 Replicas: 0,1,2 Isr: 1,2,0 Topic: four Partition: 2 Leader: 0 Replicas: 0,1,2 Isr: 0,1,2 [dhapp@conch01 kafka_2.12-3.0.0]$
八、文件存储机制
1）Topic 数据的存储机制
2）思考：Topic 数据到底存储在什么位置？
（1）启动生产者，并发送消息。[dhapp@conch01 kafka_2.12-3.0.0]$ bin/kafka-console-producer.sh -- bootstrap-server conch01:9092 --topic first >hello world（2）查看 conch01（或者 conch02、conch03）的/home/dhapp/software/kafka_2.12-3.0.0/kafka-logs/first-1（first-0、first-2）路径上的文件。
[dhapp@conch01 first-1]$ ll total 20 -rw-rw-r--. 1 dhapp dhapp 10485760 Apr 8 20:48 00000000000000000000.index -rw-rw-r--. 1 dhapp dhapp 1459 Apr 9 20:06 00000000000000000000.log -rw-rw-r--. 1 dhapp dhapp 10485756 Apr 8 20:48 00000000000000000000.timeindex -rw-rw-r--. 1 dhapp dhapp 10 Apr 7 23:19 00000000000000000032.snapshot -rw-rw-r--. 1 dhapp dhapp 14 Apr 9 18:08 leader-epoch-checkpoint -rw-rw-r--. 1 dhapp dhapp 43 Apr 7 20:49 partition.metadata [dhapp@conch01 first-1]$（3）直接查看 log 日志，发现是乱码。
[dhapp@conch01 first-1]$ cat 00000000000000000000.log ;U▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒aaa;▒Շ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒aaam▒y/▒▒!7▒▒!7▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ test0 test1 test2 test3 -▒*]▒▒*]▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒."test测试回调0."test测试回调1."test测试回调2."test测试回调3"test测试回调4 ▒L▒▒l▒▒▒l▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒RFtest测试指定分区没有key值0RFtest测试指定分区没有key值1RFtest测试指定分区没有key值2RFtest测试指定分区没有key值3Ftest测试指定分区没有key值z▒▒ ▒o.a▒o.q▒▒▒▒▒▒▒▒▒▒▒▒▒▒TaFtest测试指定分区没有key值0T aFtest测试指定分区没有key值1T aFtest测试指定分区没有key值2T aFtest测试指定分区没有key值3TaFtest测试指定分区没有key值4▒J[v▒zYR▒zY]▒▒▒▒▒▒▒▒▒▒▒▒▒▒@4test测试自定义分区0@4test测试自定义分区1@4test测试自定义分区2@4test测试自定义分区34test测试自定义分区4 hello[dhapp@conch01 first-1]$ xterm-256colorxterm-256colorxterm-256colorxterm-256colorxterm-256color（4）通过工具查看 index 和 log 信息。
[dhapp@conch01 first-1]$ kafka-run-class.sh kafka.tools.DumpLogSegments --files ./00000000000000000000.index Dumping ./00000000000000000000.index offset: 0 position: 0 [dhapp@conch01 first-1]$[dhapp@conch01 first-1]$ kafka-run-class.sh kafka.tools.DumpLogSegments --files ./00000000000000000000.log Dumping ./00000000000000000000.log Starting offset: 0 baseOffset: 0 lastOffset: 0 count: 1 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 0 CreateTime: 1649336338617 size: 71 magic: 2 compresscodec: none crc: 1436348543 isvalid: true baseOffset: 1 lastOffset: 1 count: 1 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 71 CreateTime: 1649336375070 size: 71 magic: 2 compresscodec: none crc: 3738535845 isvalid: true baseOffset: 2 lastOffset: 6 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 142 CreateTime: 1649336727461 size: 121 magic: 2 compresscodec: none crc: 2289643454 isvalid: true baseOffset: 7 lastOffset: 11 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 263 CreateTime: 1649337327079 size: 181 magic: 2 compresscodec: none crc: 4196928813 isvalid: true baseOffset: 12 lastOffset: 16 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 444 CreateTime: 1649341668601 size: 271 magic: 2 compresscodec: none crc: 4246228217 isvalid: true baseOffset: 17 lastOffset: 21 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 715 CreateTime: 1649341836913 size: 276 magic: 2 compresscodec: none crc: 2058921226 isvalid: true baseOffset: 22 lastOffset: 26 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 991 CreateTime: 1649342568797 size: 226 magic: 2 compresscodec: none crc: 122313590 isvalid: true baseOffset: 27 lastOffset: 31 count: 5 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 0 isTransactional: false isControl: false position: 1217 CreateTime: 1649342899451 size: 169 magic: 2 compresscodec: snappy crc: 2967939924 isvalid: true baseOffset: 32 lastOffset: 32 count: 1 baseSequence: -1 lastSequence: -1 producerId: -1 producerEpoch: -1 partitionLeaderEpoch: 12 isTransactional: false isControl: false position: 1386 CreateTime: 1649505958878 size: 73 magic: 2 compresscodec: none crc: 3745951668 isvalid: true [dhapp@conch01 first-1]$3）index 文件和 log 文件详解
说明：日志存储参数配置
参数 描述
log.segment.bytes Kafka 中 log 日志是分成一块块存储的，此配置是指 log 日志划分成块的大小，默认值 1G。
log.index.interval.bytes 默认 4kb，kafka 里面每当写入了 4kb 大小的日志（.log），然后就往 index 文件里面记录一个索引。 稀疏索引。
九、文件清理策略
Kafka 中默认的日志保存时间为 7 天，可以通过调整如下参数修改保存时间。
⚫ log.retention.hours，最低优先级小时，默认 7 天。
⚫ log.retention.minutes，分钟。
⚫ log.retention.ms，最高优先级毫秒。
⚫ log.retention.check.interval.ms，负责设置检查周期，默认 5 分钟。那么日志一旦超过了设置的时间，怎么处理呢？
Kafka 中提供的日志清理策略有delete 和 compact 两种。1）delete 日志删除：将过期数据删除
⚫ log.cleanup.policy = delete 所有数据启用删除策略
（1）基于时间：默认打开。以 segment 中所有记录中的最大时间戳作为该文件时间戳。
（2）基于大小：默认关闭。超过设置的所有日志总大小，删除最早的 segment。log.retention.bytes，默认等于-1，表示无穷大。思考：如果一个 segment 中有一部分数据过期，一部分没有过期，怎么处理？
2）compact 日志压缩
十、高效读写数据
1）Kafka 本身是分布式集群，可以采用分区技术，并行度高
提高生产端和消费端并行度，同时可以把海量的数据打散。
2）读数据采用稀疏索引，可以快速定位要消费的数据
3）顺序写磁盘Kafka 的 producer 生产数据，要写入到 log 文件中，写的过程是一直追加到文件末端，为顺序写。官网有数据表明，同样的磁盘，顺序写能到 600M/s，而随机写只有 100K/s。这与磁盘的机械机构有关，顺序写之所以快，是因为其省去了大量磁头寻址的时间。
4）页缓存 + 零拷贝技术
主要的是在集群中不处理数据，处理数据主要在生产端和消费端的拦截器，序列化和反序列化中进行。
log.flush.interval.messages 强制页缓存刷写到磁盘的条数，默认是 long 的最大值，9223372036854775807。一般不建议修改，交给系统自己管理。
log.flush.interval.ms 每隔多久，刷数据到磁盘，默认是 null。一般不建议修改，交给系统自己管理。
2712
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
