Kafka集群擴展以及重新分布分區

我們往已經部署好的Kafka集群里面添加機器是最正常不過的需求，而且添加起來非常地方便，我們需要做的事是從已經部署好的Kafka節點中復制相應的配置文件，然后把里面的broker id修改成全局唯一的，最后啟動這個節點即可將它加入到現有Kafka集群中。

　　但是問題來了，新添加的Kafka節點並不會自動地分配數據，所以無法分擔集群的負載，除非我們新建一個topic。但是現在我們想手動將部分分區移到新添加的Kafka節點上，Kafka內部提供了相關的工具來重新分布某個topic的分區。在重新分布topic分區之前，我們先來看看現在topic的各個分區的分布位置：

. /bin/kafka-topics .sh --topic iteblog --describe --zookeeper www.iteblog.com:2181

Topic:iteblog PartitionCount:7  ReplicationFactor:2 Configs:

   Topic: iteblog  Partition: 0  Leader: 1 Replicas: 1,2 Isr: 1,2

   Topic: iteblog  Partition: 1  Leader: 2 Replicas: 2,3 Isr: 2,3

   Topic: iteblog  Partition: 2  Leader: 3 Replicas: 3,4 Isr: 3,4

   Topic: iteblog  Partition: 3  Leader: 4 Replicas: 4,1 Isr: 4,1

   Topic: iteblog  Partition: 4  Leader: 1 Replicas: 1,3 Isr: 1,3

   Topic: iteblog  Partition: 5  Leader: 2 Replicas: 2,4 Isr: 2,4

   Topic: iteblog  Partition: 6  Leader: 3 Replicas: 3,1 Isr: 3,1

從上面的輸出可以看出，iteblog主題一共有7個分區，但是我們broker的個數只有4個，所以會導致某些broker維護更多的分區。現在我們在現有集群的基礎上再添加一個Kafka節點，然后使用Kafka自帶的kafka-reassign-partitions.sh工具來重新分布分區。該工具有三種使用模式：

　　1、generate模式，給定需要重新分配的Topic，自動生成reassign plan（並不執行）
　　2、execute模式，根據指定的reassign plan重新分配Partition
　　3、verify模式，驗證重新分配Partition是否成功

現在我們需要將原先分布在broker 1-4節點上的分區重新分布到broker 1-5節點上，借助kafka-reassign-partitions.sh工具生成reassign plan，不過我們先得按照要求定義一個文件，里面說明哪些topic需要重新分區，文件內容如下：

[[e-mail] ~]$ cat topics-to-move.json

{ "topics" : [{ "topic" : "iteblog" }],

  "version" :1

}

然后使用kafka-reassign-partitions.sh工具生成reassign plan

[[e-mail] ~]$ bin/kafka-reassign-partitions.sh --zookeeper www.iteblog.com:2181 --topics-to-move-json-file topics-to-move.json --broker-list "1,2,3,4,5" --generate Current partition replica assignment {"version":1,"partitions":[{"topic":"iteblog","partition":3,"replicas":[4,1]},{"topic":"iteblog","partition":5,"replicas":[2,4]},{"topic":"iteblog","partition":4,"replicas":[1,3]},{"topic":"iteblog","partition":0,"replicas":[1,2]},{"topic":"iteblog","partition":6,"replicas":[3,1]},{"topic":"iteblog","partition":1,"replicas":[2,3]},{"topic":"iteblog","partition":2,"replicas":[3,4]}]} Proposed partition reassignment configuration {"version":1,"partitions":[{"topic":"iteblog","partition":3,"replicas":[3,5]},{"topic":"iteblog","partition":5,"replicas":[5,3]},{"topic":"iteblog","partition":4,"replicas":[4,1]},{"topic":"iteblog","partition":0,"replicas":[5,2]},{"topic":"iteblog","partition":6,"replicas":[1,4]},{"topic":"iteblog","partition":1,"replicas":[1,3]},{"topic":"iteblog","partition":2,"replicas":[2,4]}]} 

Proposed partition reassignment configuration下面生成的就是將分區重新分布到broker 1-5上的結果。我們將這些內容保存到名為result.json文件里面（文件名不重要，文件格式也不一定要以json為結尾，只要保證內容是json即可），然后執行這些reassign plan：

[[e-mail] ~]$ bin/kafka-reassign-partitions.sh --zookeeper www.iteblog.com:2181 --reassignment-json-file result.json --execute Current partition replica assignment {"version":1,"partitions":[{"topic":"iteblog","partition":3,"replicas":[4,1]},{"topic":"iteblog","partition":5,"replicas":[2,4]},{"topic":"iteblog","partition":4,"replicas":[1,3]},{"topic":"iteblog","partition":0,"replicas":[1,2]},{"topic":"iteblog","partition":6,"replicas":[3,1]},{"topic":"iteblog","partition":1,"replicas":[2,3]},{"topic":"iteblog","partition":2,"replicas":[3,4]}]} Save this to use as the --reassignment-json-file option during rollback Successfully started reassignment of partitions {"version":1,"partitions":[{"topic":"iteblog","partition":1,"replicas":[1,3]},{"topic":"iteblog","partition":5,"replicas":[5,3]},{"topic":"iteblog","partition":4,"replicas":[4,1]},{"topic":"iteblog","partition":6,"replicas":[1,4]},{"topic":"iteblog","partition":2,"replicas":[2,4]},{"topic":"iteblog","partition":0,"replicas":[5,2]},{"topic":"iteblog","partition":3,"replicas":[3,5]}]} 

這樣Kafka就在執行reassign plan，我們可以校驗reassign plan是否執行完成：

[[e-mail] ~]$ bin/kafka-reassign-partitions.sh --zookeeper www.iteblog.com:2181 --reassignment-json-file result.json --verify Status of partition reassignment: Reassignment of partition [iteblog,1] completed successfully Reassignment of partition [iteblog,5] is still in progress Reassignment of partition [iteblog,4] completed successfully Reassignment of partition [iteblog,6] completed successfully Reassignment of partition [iteblog,2] completed successfully Reassignment of partition [iteblog,0] is still in progress Reassignment of partition [iteblog,3] completed successfully [[e-mail] ~]$ bin/kafka-reassign-partitions.sh --zookeeper www.iteblog.com:2181 --reassignment-json-file result.json --verify Status of partition reassignment: Reassignment of partition [iteblog,1] completed successfully Reassignment of partition [iteblog,5] completed successfully Reassignment of partition [iteblog,4] completed successfully Reassignment of partition [iteblog,6] completed successfully Reassignment of partition [iteblog,2] completed successfully Reassignment of partition [iteblog,0] completed successfully Reassignment of partition [iteblog,3] completed successfully 

可以看出，分區正在Reassignment的狀態是still in progress；如果分區Reassignment完成則completed successfully，然后我們就可以看到分區已經按照生成的reassign plan進行，我們可以看下topic各個分區現在的分布情況：

[[e-mail] ~]$ ./bin/kafka-topics.sh --topic iteblog --describe --zookeeper www.iteblog.com:2181 Topic:iteblog PartitionCount:7 ReplicationFactor:2 Configs: Topic: iteblog Partition: 0 Leader: 5 Replicas: 5,2 Isr: 2,5 Topic: iteblog Partition: 1 Leader: 1 Replicas: 1,3 Isr: 3,1 Topic: iteblog Partition: 2 Leader: 2 Replicas: 2,4 Isr: 4,2 Topic: iteblog Partition: 3 Leader: 3 Replicas: 3,5 Isr: 3,5 Topic: iteblog Partition: 4 Leader: 1 Replicas: 4,1 Isr: 1,4 Topic: iteblog Partition: 5 Leader: 5 Replicas: 5,3 Isr: 3,5 Topic: iteblog Partition: 6 Leader: 1 Replicas: 1,4 Isr: 1,4 

分區的分布的確和操作之前不一樣了，broker 5上已經有分區分布上去了。但是仔細的同學應該可以發現，broker 4上居然沒有分區的Leader，這肯定不是我們想要的！所以使用kafka-reassign-partitions.sh工具生成的reassign plan只是一個建議，方便大家而已。其實我們自己完全可以編輯一個reassign plan，然后執行它，如下：

{

     "version" : 1,

     "partitions" : [

         {

             "topic" : "iteblog" ,

             "partition" : 0,

             "replicas" : [

                 1,

                 2

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 1,

             "replicas" : [

                 2,

                 3

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 2,

             "replicas" : [

                 3,

                 4

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 3,

             "replicas" : [

                 4,

                 5

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 4,

             "replicas" : [

                 5,

                 1

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 5,

             "replicas" : [

                 1,

                 3

             ]

         },

         {

             "topic" : "iteblog" ,

             "partition" : 6,

             "replicas" : [

                 2,

                 4

             ]

         }

     ]

}

將上面的json數據文件保存到result.json文件中，然后也是執行它：

[[e-mail] ~]$ bin/kafka-reassign-partitions.sh --zookeeper www.iteblog.com:2181 --reassignment-json-file result.json --execute 

等這個reassign plan執行完，我們再來看看分區的分布：

[[e-mail] ~]$ ./bin/kafka-topics.sh --topic iteblog --describe --zookeeper www.iteblog.com:2181 Topic:iteblog PartitionCount:7 ReplicationFactor:2 Configs: Topic: iteblog Partition: 0 Leader: 1 Replicas: 1,2 Isr: 2,1 Topic: iteblog Partition: 1 Leader: 2 Replicas: 2,3 Isr: 3,2 Topic: iteblog Partition: 2 Leader: 3 Replicas: 3,4 Isr: 4,3 Topic: iteblog Partition: 3 Leader: 4 Replicas: 4,5 Isr: 5,4 Topic: iteblog Partition: 4 Leader: 5 Replicas: 5,1 Isr: 1,5 Topic: iteblog Partition: 5 Leader: 1 Replicas: 1,3 Isr: 3,1 Topic: iteblog Partition: 6 Leader: 2 Replicas: 2,4 Isr: 4,2 

果然已經按照我們需求分布了。。