黑猴子的家：Zookeeper 在线迁移

于 2018-09-27 15:24:00 发布
阅读量379 收藏 2
点赞数
分类专栏： Zookeeper
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/qq_28652401/article/details/83508142
版权

在至少有一个Leader存在的前提下，进行Zookeeper的在线增量、在线减量、在线迁移
在全过程中ZooKeeper不停止服务

1、 注意事项

首先，当我们要从3台扩充到5台时，应保证集群不停止服务。
3台不停止服务的最低限度是2台（X/2+1），而5台的最低限度是3台。
我们应该保证，集群中最低有3台ZooKeeper是启动的。
此外，重启时应保证先重启myid最小的机器，由小向大进行重启
Leader无论其myid大小，都放到最后重启
因为ZooKeeper的机制中，myid大的会向小的发起连接，而小的不会向大的发起连接。因此如果最后重启myid最小的机器，则其可能无法加入集群

2、环境情况

五台机器

IPHostname
192.168.2.102hadoop102
192.168.2.103hadoop103
192.168.2.104hadoop104
192.168.2.105hadoop105
192.168.2.106hadoop106

jdk

1.8.144

zookeeper

zookeeper-3.4.10

myid

1-5

配置文件

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888

3、 配置一个3节点的ZooKeeper

hadoop102

[victor@hadoop102 bin]$./zkServer.shstatus
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.10/bin/../conf/zoo.cfg
Mode:follower

hadoop103

[victor@hadoop103 bin]$ ./zkServer.sh status
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.10/bin/../conf/zoo.cfg
Mode: leader

hadoop104

[victor@hadoop104 bin]$./zkServer.shstatus
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.10/bin/../conf/zoo.cfg
Mode:follower

4、将其扩容为5节点的ZooKeeper

1）先查看原先的ZooKeeper集群情况

echo mntr|nc localhost 2181
这条4字命令可以查看集群的情况，其中follower的相关数据需要在Leader机器上才能查看

2）在hadoop103上查看

[victor@hadoop103 bin]$ echo mntr|nc localhost 2181
zk_version  3.4.6-1569965, built on 02/20/2014 09:09 GMT
zk_avg_latency  0
zk_max_latency  0
zk_min_latency  0
zk_packets_received 3
zk_packets_sent 2
zk_num_alive_connections    1
zk_outstanding_requests 0
zk_server_state leaderzk_znode_count  4
zk_watch_count  0zk_ephemerals_count 0
zk_approximate_data_size    27
zk_open_file_descriptor_count   27
zk_max_file_descriptor_count    65535
zk_followers    2zk_synced_followers 2
zk_pending_syncs    0

3）启动另外两台机器的Zookeeper，另外两台机器的配置文件

zoo.cfg

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

4）启动

hadoop105

[victor@hadoop105 bin]# ./zkServer.sh status
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode: follower

hadoop106

[victor@hadoop106 bin]# ./zkServer.sh status
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:follower

5）再查看集群情况

仍然在hadoop103上查看

[victor@hadoop103 bin]$ echo mntr|nc localhost 2181
zk_version  3.4.6-1569965, built on 02/20/2014 09:09 GMT
zk_avg_latency  0
zk_max_latency  0
zk_min_latency  0
zk_packets_received 4
zk_packets_sent 3
zk_num_alive_connections    1
zk_outstanding_requests 0
zk_server_state leader
zk_znode_count  4
zk_watch_count  0
zk_ephemerals_count 0
zk_approximate_data_size    27
zk_open_file_descriptor_count   31
zk_max_file_descriptor_count    65535
zk_followers    4
zk_synced_followers 4
zk_pending_syncs    0

可以看到zk_followers为4，连接到的follower从2变为4了,而且zk_synced_followers为4，说明新加入的2个也都同步好了,接下来我们滚动重启myid为1-3的前三台机器先处理hadoop102关闭,如不放心请在关闭其间于Leader机器或后加入的两台机器上监控日志

6）关闭 hadoop102

[victor@hadoop102 bin]$./zkServer.shstop
JMXenabledbydefault
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stoppingzookeeper...STOPPED

7）修改其配置文件

由原来的

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888

到新的

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

启动

[victor@hadoop102 bin]$ ./zkServer.sh start
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED

[victor@hadoop102 bin]$ ./zkServer.sh status
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode: follower 

然后跳过作为Leader的hadoop103，先处理hadoop104

8）hadoop104关闭

[victor@hadoop104bin]$./zkServer.shstop
JMXenabledbydefault
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stoppingzookeeper...STOPPED

修改其配置文件

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

启动

[victor@hadoop104bin]$./zkServer.sh start
JMX enabledby default
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Startingzookeeper...STARTED

[victor@hadoop104bin]$./zkServer.sh status
JMX enabledby default
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:follower

9）最后处理原Leader的hadoop103

关闭

[victor@hadoop103 bin]$ ./zkServer.sh stop
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stopping zookeeper ... STOPPED

新ZooKeeper Leader
查看新Leader ZooKeeper会尽可能的选择myid最大的机器为Leader，因此原本的hadoop106其myid为5变为了Leader

[victor@hadoop106 bin]# ./zkServer.sh status
JMXenabledbydefault
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:leader

hadoop103修改配置文件

server.1=hadoop102:2888:3888
server.2=hadoop103:2888:3888
server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

启动

[victor@hadoop103 bin]$./zkServer.sh start
JMX enabledby default
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Startingzookeeper...STARTED

[victor@hadoop103bin]$./zkServer.sh status
JMX enabledby default
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:follower

在新的Leader上hadoop106查看集群情况

[victor@hadoop106 bin]# echo mntr|nc localhost 2181
zk_version  3.4.6-1569965, built on 02/20/2014 09:09 GMT
zk_avg_latency  1
zk_max_latency  4
zk_min_latency  0
zk_packets_received 12
zk_packets_sent 11
zk_num_alive_connections    1
zk_outstanding_requests 0
zk_server_state leader
zk_znode_count  4
zk_watch_count  0
zk_ephemerals_count 0
zk_approximate_data_size    27
zk_open_file_descriptor_count   33
zk_max_file_descriptor_count    65535
zk_followers    4zk_synced_followers 4
zk_pending_syncs    0

一切正常到这里，我们已经将原本的3台扩展到了5台，成功了一半。然后只要将现在的5台再缩小到3台,且不包括原本myid为1-2的机器，就完成了迁移,将5台缩小回3台修改hadoop104根据前面的注意事项，我们此时5台集群中启动的数量不得少于3台，因此我们需要先修改3-5号机器的配置文件为3台，再关闭1-2号机器关闭

5、将其缩小为3台

1） hadoop104关闭

[victor@hadoop104 bin]$./zkServer.sh stop
JMX enabledby default
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stoppingzookeeper...STOPPED

2）修改hadoop104配置文件为

server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

3）hadoop104启动

[victor@hadoop104 bin]$./zkServer.sh start
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Starting zookeeper...STARTED

[victor@hadoop104 bin]$./zkServer.sh status
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:follower

4）然后修改hadoop105

关闭

[victor@hadoop105 bin]# ./zkServer.sh stop
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stopping zookeeper ...

5）hadoop105 stop 修改配置文件为

server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

6）启动hadoop105

[victor@hadoop105 bin]$ ./zkServer.sh start
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED

[victor@hadoop105 bin]$ ./zkServer.sh status
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfgMode: follower

7）最后修改hadoop106

关闭hadoop106

[victor@hadoop106bin]$./zkServer.shstop
JMXenabledbydefault
Usingconfig:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stoppingzookeeper...STOPPED

8）修改hadoop106配置文件为

server.3=hadoop104:2888:3888
server.4=hadoop105:2888:3888
server.5=hadoop106:2888:3888

关闭后Leader移动到了myid第二大的hadoop105上

9）启动

[victor@hadoop106 bin]$./zkServer.sh start
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Starting zookeeper...STARTED

[victor@hadoop106 bin]$./zkServer.sh status
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Mode:follower

10）在hadoop105 Leader中查看

[victor@hadoop105 bin]$ echo mntr| nc localhost  2181
zk_version  3.4.6-1569965, built on 02/20/2014 09:09 GMT
zk_avg_latency  0
zk_max_latency  0
zk_min_latency  0
zk_packets_received 4
zk_packets_sent 3
zk_num_alive_connections    1
zk_outstanding_requests 0
zk_server_state leader
zk_znode_count  4
zk_watch_count  0
zk_ephemerals_count 0
zk_approximate_data_size    27
zk_open_file_descriptor_count   27
zk_max_file_descriptor_count    65535
zk_followers    2zk_synced_followers 2
zk_pending_syncs    0

此时的zk_followers为2，说明Leader已经不认1-2号机器了,关闭1-2号机器

11）关闭hadoop102

[victor@hadoop102 bin]$./zkServer.shstop
JMX enabledby default
Using config:/usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stopping zookeeper...STOPPED

12）关闭hadoop103

[victor@hadoop103 bin]$ ./zkServer.sh stop
JMX enabled by default
Using config: /usr/local/webserver/zookeeper-3.4.6/bin/../conf/zoo.cfg
Stopping zookeeper ...

13）hadoop105 Leader 再查看

[victor@hadoop105 bin]$ echo mntr | nclocalhost 2181
zk_version  3.4.6-1569965,builton02/20/201409:09GMT
zk_avg_latency  0
zk_max_latency  0
zk_min_latency  0
zk_packets_received5
zk_packets_sent4
zk_num_alive_connections    1
zk_outstanding_requests0
zk_server_stateleader
zk_znode_count  4
zk_watch_count  0
zk_ephemerals_count0
zk_approximate_data_size    27
zk_open_file_descriptor_count  27
zk_max_file_descriptor_count    65535
zk_followers    2
zk_synced_followers2
zk_pending_syncs    0
黑猴子的家
关注
专栏目录
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值