redis 集群安装

redis集群安装

1.下载redis源码

2.解压并进入解压后的文件夹redis内

3.make,生成一系列的文件(mkreleasehdr.sh, redis-benchmark, redis-check-aof, redis-check-rdb, redis-cli, redis-sentinel, redis-server, redis-server, redis-trib.rb)

要部署集群,至少需要6个节点,这里使用了六台电脑centos系统,每台电脑上运行一个redis实例,端口为6379 。(如果没有6台,则可以在一台电脑上运行多个redis实例,设置不同的端口即可)

4.设置配置文件 src/redis.conf

port 6379  # 绑定端口,可以改为其他端口

daemon yes   # 后台运行

cluster-enabled yes #集群
cluster-config-file nodes.conf  # 组成集群时自动生成的配置文件,用户不需修改这个文件
cluster-node-timeout 5000    # 超时(认为宕机)
appendonly yes          # 每次更新或插入新数据时写入日志文件

bind

bind <本机ip>  # 绑定ip,目的是为了让其他节点知道,如果在一台电脑上运行6个redis实例,则这里可以设为127.0.0.1(没有实测)

5. 安装ruby,rubygems,可以从网上下载源码安装或者 yum install

6. 执行指令

gem install redis

这是因为集群启动是使用ruby脚本,如果出现安装错误比如说没有zlib或者openssl,则需要yum install zlib zlib-devel openssl openssl-devel,然后需要重新编译安装ruby(否则仍然提示错误)。

7. 将redis文件夹复制到其他5台电脑上,并在6台电脑上依次启动6个redis实例

8. 执行指令

./redis-trib.rb create --replicas 1 127.0.0.1:7000 127.0.0.1:7001 \
127.0.0.1:7002 127.0.0.1:7003 127.0.0.1:7004 127.0.0.1:7005

将127.0.0.1:<port> 替换为实际的ip:port

如果出现redis waiting for the cluster to join.... 一直阻塞,则说明有的节点端口无法通信,可以开放所绑定的端口(以及 绑定端口+10000 得到的另一个端口,这个端口redis也使用)

如果出现 sorry, can't connect to node xxx.xxx.xx.xx,则很可能是防火墙没有开放端口

进行如下设置即可

开启端口
 
firewall-cmd --zone=public --add-port=6379/tcp --permanent
 firewall-cmd --zone=public --add-port=16379/tcp --permanent
命令含义:
 
--zone #作用域
 
--add-port=80/tcp  #添加端口,格式为:端口/通讯协议
 
--permanent   #永久生效,没有此参数重启后失效
 
重启防火墙
 
firewall-cmd --reload

 

9.如果需要关闭集群,则需要在每台电脑上分别关闭

ps -aux | grep redis

得到redis的pid,然后

kill -9 <pid>

此时如果再次执行上面 7,8 步骤重启集群,则会出现类似

Node 192.168.10.219:6379 is not empty. Either the node already knows other nodes (check with CLUSTER NODES) or contains some key in database 0.

的错误,此时需要将redis启动目录中的appendonly.aof,nodes.conf,dump.rdb文件删除后再执行 7,8步骤方可

 

10. 最后登录集群

./redis-cli -h 192.168.10.10 -c -p 6379

11. 如果给集群设置密码,则注意在集群搭建完成前不要设置密码,在搭建之后通过客户端设置密码,如下(注意分别登录每个节点并设置)

config set masterauth <p1>
config set requirepass <p2>
auth <p2>
config rewrite

下次登录可使用

./redis-cli -h <ip> -c -p 6379 -a <p2>

或者使用10步骤登录后,再使用 auth <p2>验证身份

 

redis集群管理工具

(以下内容来源:http://blog.sina.com.cn/s/blog_13ea37b210102wyc7.html

redis-trib.rb是redis官方推出的管理redis集群的工具,集成在redis的源码src目录下,是基于redis提供的集群命令封装成简单、便捷、实用的操作工具。redis-trib.rb是redis作者用ruby完成的。

先从redis-trib.rb的help信息,看下redis-trib.rb提供了哪些功能。
redis-trib
选项:
  create          host1:port1 ... hostN:portN  创建集群,只有大于等于3个节点才能创建集群
                      --replicas                              每个节点需要几个slave
  check           host:port                              检查集群,只需要选择一个集群中的一个节点即可
  info             host:port                               查看集群信息,只需要选择一个集群中的一个节点即可
  fix               host:port                               修复集群,目前fix命令能修复两种异常,一种是集群有处
                                                                  于迁移中的slot的节点,一种是slot未完全分配的异常
                     --timeout                              超时时间
  reshard         host:port                             在线迁移slot
                    --from                   需要从哪些源节点上迁移slot,可从多个源节点完成迁移,
                                                  以逗号隔开,传递的是节点的node id,
                                                  还可以直接传递--from all,这样源节点就是集群的所有节点,
                                                  不传递该参数的话,则会在迁移过程中提示用户输入。
                  --to                    Slot需要迁移的目的节点的node id,目的节点只能填写一个,
                                            不传递该参数的话,则会在迁移过程中提示用户输入。
                  --slots                 需要迁移的slot数量,不传递该参数的话,则会在迁移过程中提示用户输入。
                  --yes                        设置该参数,可以在打印执行reshard计划的时候,提示用户输入yes确认后再执行reshard
                  --timeout               设置migrate命令的超时时间。
                  --pipeline              定义cluster getkeysinslot命令一次取出的key数量,
                                                不传的话使用默认值为10。
  rebalance       host:port         平衡集群节点slot数量,这个是必传参数,
                                                用来从一个节点获取整个集群信息,相当于获取集群信息的入口
                  --weight                节点的权重,格式为node_id=weight,如果需要为多个节点分配权
                                                重的话,需要添加多个--weight 参数,
                                                即--weight b31e3a2e=5 --weight 60b8e3a1=5,node_id可为节点
                                                名称的前缀,只要保证前缀位数能唯一区分该节点即可。
                                                没有传递–weight的节点的权重默认为1。
                  --auto-weights               这个参数在rebalance流程中并未用到
                  --use-empty-masters          rebalance是否考虑没有节点的master,
                                               默认没有分配slot节点的master是不参与rebalance的,
                                               设置--use-empty-masters可以让没有分配slot的节点参与rebalance
                  --timeout              设置migrate命令的超时时间
                  --simulate             设置该参数,可以模拟rebalance操作,提示用户会迁移哪些slots,
                                                而不会真正执行迁移操作
                  --pipeline              与reshar的pipeline参数一样,定义cluster getkeysinslot命令一次
                                                取出的key数量,不传的话使用默认值为10。
                  --threshold             只有节点需要迁移的slot阈值超过threshold,才会执行rebalance操作
  add-node        new_host:new_port existing_host:existing_port 将新节点加入集群,
                                               命令可以将新节点加入集群,节点可以为master,也可
                                               以为某个master节点的slave
                  --slave                  设置该参数,则新节点以slave的角色加入集群
                  --master-id           这个参数需要设置了--slave才能生效,--master-id用来指定新节点的
                                               master节点。如果不设置该参数,则会随机为节点选择master节点。
  del-node        host:port node_id            从集群中删除节点,
  set-timeout     host:port milliseconds       设置集群节点间心跳连接的超时时间,单位是毫秒
  call            host:port command arg arg .. arg  在集群全部节点上执行命令,
                                              call命令也是需要通过集群的一个节点地址,连上整个集群,
                                              然后在集群的每个节点执行该命令。
  import          host:port        将外部redis节点数据导入集群,
                                              import命令更适合离线的把外部redis数据导入,
                                               以slave的方式连接redis节点去同步节点数据应该是更好的方式。
                  --from                  源节点
                  --copy                       拷贝
                  --replace                    替换
  help            打印帮助信息
 
示例:
* 创建Redis集群
$ redis-trib.rb create --replicas 1 172.16.0.7:6379 172.16.0.7:6380 172.16.0.7:6381 172.16.0.7:6382 172.16.0.7:6383 172.16.0.7:6384
>>> Creating cluster
>>> Performing hash slots allocation on 6 nodes...
Using 3 masters:
172.16.0.7:6379
172.16.0.7:6380
172.16.0.7:6381
Adding replica 172.16.0.7:6382 to 172.16.0.7:6379
Adding replica 172.16.0.7:6383 to 172.16.0.7:6380
Adding replica 172.16.0.7:6384 to 172.16.0.7:6381
M: 8b3809e6e5591dcfec067216906a25d7cd482598 172.16.0.7:6379
   slots:0-5460 (5461 slots) master
M: a1ac1fc64549ca15eb889395c46d001162bf2aa6 172.16.0.7:6380
   slots:5461-10922 (5462 slots) master
M: 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5 172.16.0.7:6381
   slots:10923-16383 (5461 slots) master
S: 7d033c0df5b7db4f872afa94566fb061c4ebe16a 172.16.0.7:6382
   replicates 8b3809e6e5591dcfec067216906a25d7cd482598
S: 72b5baa2dbdb315fef604ea8b6deb0143372ae1b 172.16.0.7:6383
   replicates a1ac1fc64549ca15eb889395c46d001162bf2aa6
S: 8641d66fce1f5cc982ef7649d53a1eba2fa9b7c5 172.16.0.7:6384
   replicates 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5
Can I set the above configuration? (type 'yes' to accept): yes
>>> Nodes configuration updated
>>> Assign a different config epoch to each node
>>> Sending CLUSTER MEET messages to join the cluster
Waiting for the cluster to join......
>>> Performing Cluster Check (using node 172.16.0.7:6379)
M: 8b3809e6e5591dcfec067216906a25d7cd482598 172.16.0.7:6379
   slots:0-5460 (5461 slots) master
M: a1ac1fc64549ca15eb889395c46d001162bf2aa6 172.16.0.7:6380
   slots:5461-10922 (5462 slots) master
M: 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5 172.16.0.7:6381
   slots:10923-16383 (5461 slots) master
M: 7d033c0df5b7db4f872afa94566fb061c4ebe16a 172.16.0.7:6382
   slots: (0 slots) master
   replicates 8b3809e6e5591dcfec067216906a25d7cd482598
M: 72b5baa2dbdb315fef604ea8b6deb0143372ae1b 172.16.0.7:6383
   slots: (0 slots) master
   replicates a1ac1fc64549ca15eb889395c46d001162bf2aa6
M: 8641d66fce1f5cc982ef7649d53a1eba2fa9b7c5 172.16.0.7:6384
   slots: (0 slots) master
   replicates 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.

* 检查Redis集群
$ redis-trib.rb check 172.16.0.7:6379
>>> Performing Cluster Check (using node 172.16.0.7:6379)
M: 8b3809e6e5591dcfec067216906a25d7cd482598 172.16.0.7:6379
   slots:0-5460 (5461 slots) master
   1 additional replica(s)
S: 8641d66fce1f5cc982ef7649d53a1eba2fa9b7c5 172.16.0.7:6384
   slots: (0 slots) slave
   replicates 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5
S: 7d033c0df5b7db4f872afa94566fb061c4ebe16a 172.16.0.7:6382
   slots: (0 slots) slave
   replicates 8b3809e6e5591dcfec067216906a25d7cd482598
M: a1ac1fc64549ca15eb889395c46d001162bf2aa6 172.16.0.7:6380
   slots:5461-10922 (5462 slots) master
   1 additional replica(s)
S: 72b5baa2dbdb315fef604ea8b6deb0143372ae1b 172.16.0.7:6383
   slots: (0 slots) slave
   replicates a1ac1fc64549ca15eb889395c46d001162bf2aa6
M: 01a0cccbb8b15e00469d1cafd2eedfb6bd8bd2b5 172.16.0.7:6381
   slots:10923-16383 (5461 slots) master
   1 additional replica(s)
[OK] All nodes agree about slots configuration.

* 查看Redis集群信息
$ redis-trib.rb info 172.16.0.7:6379
172.16.0.7:6379 (8b3809e6...) -> 0 keys | 5461 slots | 1 slaves.
172.16.0.7:6380 (a1ac1fc6...) -> 0 keys | 5462 slots | 1 slaves.
172.16.0.7:6381 (01a0cccb...) -> 0 keys | 5461 slots | 1 slaves.
[OK] 0 keys in 3 masters.
0.00 keys per slot on average.

* 添加Redis节点
$ redis-trib.rb add-node 172.16.0.7:6385 172.16.0.7:6379
>>> Adding node 172.16.0.7:6385 to cluster 172.16.0.7:6379
>>> Performing Cluster Check (using node 172.16.0.7:6379)
M: 328fe71182234df381fb4845e3f4bb4c3288320d 172.16.0.7:6379
   slots:0-5460 (5461 slots) master
   1 additional replica(s)
S: da6149e83031f8e1d790784b8650732365134dee 172.16.0.7:6383
   slots: (0 slots) slave
   replicates 237d335822d545e97723224dc965bb986586c98e
M: 6d7ad892ceb964063db08fbb6c1d7b0f6226418c 172.16.0.7:6381
   slots:10923-16383 (5461 slots) master
   1 additional replica(s)
M: 237d335822d545e97723224dc965bb986586c98e 172.16.0.7:6380
   slots:5461-10922 (5462 slots) master
   1 additional replica(s)
S: 23c093add21934a6b3eeba18897f536947c89f42 172.16.0.7:6384
   slots: (0 slots) slave
   replicates 6d7ad892ceb964063db08fbb6c1d7b0f6226418c
S: 4341c643d2b785019bc940e1ffcf5e44b2343e09 172.16.0.7:6382
   slots: (0 slots) slave
   replicates 328fe71182234df381fb4845e3f4bb4c3288320d
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Send CLUSTER MEET to node 172.16.0.7:6385 to make it join the cluster.
[OK] New node added correctly.

* 添加Redis节点
$ redis-trib.rb add-node --slave --master-id 9d6229abfea1fa5a3faa639a7b54ac3a7810f9ff 172.16.0.7:6386 172.16.0.7:6379
>>> Adding node 172.16.0.7:6386 to cluster 172.16.0.7:6379
>>> Performing Cluster Check (using node 172.16.0.7:6379)
M: 328fe71182234df381fb4845e3f4bb4c3288320d 172.16.0.7:6379
   slots:0-5460 (5461 slots) master
   1 additional replica(s)
S: da6149e83031f8e1d790784b8650732365134dee 172.16.0.7:6383
   slots: (0 slots) slave
   replicates 237d335822d545e97723224dc965bb986586c98e
M: 6d7ad892ceb964063db08fbb6c1d7b0f6226418c 172.16.0.7:6381
   slots:10923-16383 (5461 slots) master
   1 additional replica(s)
M: 9d6229abfea1fa5a3faa639a7b54ac3a7810f9ff 172.16.0.7:6385
   slots: (0 slots) master
   0 additional replica(s)
M: 237d335822d545e97723224dc965bb986586c98e 172.16.0.7:6380
   slots:5461-10922 (5462 slots) master
   1 additional replica(s)
S: 23c093add21934a6b3eeba18897f536947c89f42 172.16.0.7:6384
   slots: (0 slots) slave
   replicates 6d7ad892ceb964063db08fbb6c1d7b0f6226418c
S: 4341c643d2b785019bc940e1ffcf5e44b2343e09 172.16.0.7:6382
   slots: (0 slots) slave
   replicates 328fe71182234df381fb4845e3f4bb4c3288320d
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Send CLUSTER MEET to node 172.16.0.7:6386 to make it join the cluster.
Waiting for the cluster to join.
>>> Configure node as replica of 172.16.0.7:6385.
[OK] New node added correctly.

2.6 平衡集群节点slot数量
$ redis-trib.rb rebalance --use-empty-masters 172.16.0.7:6379
>>> Performing Cluster Check (using node 172.16.0.7:6379)
[OK] All nodes agree about slots configuration.
>>> Check for open slots...
>>> Check slots coverage...
[OK] All 16384 slots covered.
>>> Rebalancing across 4 nodes. Total weight = 4
Moving 1366 slots from 172.16.0.7:6380 to 172.16.0.7:6385
##########################################################################
Moving 1365 slots from 172.16.0.7:6381 to 172.16.0.7:6385
##########################################################################
Moving 1365 slots from 172.16.0.7:6379 to 172.16.0.7:6385
##########################################################################

* 在线迁移Slot
$ redis-trib.rb reshard --from 9d6229abfea1fa5a3faa639a7b54ac3a7810f9ff --to 328fe71182234df381fb4845e3f4bb4c3288320d --slots 4096 172.16.0.7:6379

* 从集群中删除节点
$ redis-trib.rb del-node 172.16.0.7:6379 9d6229abfea1fa5a3faa639a7b54ac3a7810f9ff
>>> Removing node 9d6229abfea1fa5a3faa639a7b54ac3a7810f9ff from cluster 172.16.0.7:6379
>>> Sending CLUSTER FORGET messages to the cluster...
>>> SHUTDOWN the node.

注意:必须确保数据迁移后,才能从集群中删除节点,否则报以下错误:
[ERR] Node 172.16.0.7:6385 is not empty! Reshard data away and try again.

posted @ 2017-02-22 19:30  gaoshoufenmu  阅读(935)  评论(0编辑  收藏  举报