【Redis】redis-cluster到redis-cluster的快速复制

现有个需求是redis-cluster到redis-cluster的快速迁移，并且迁移过后两套redis-cluster都要独立使用无依赖

方法一：使用前面redis迁移中的在线迁移工具

redis-migration（劣势：不支持redis4.0）redis-migration-tools（劣势：不支持redis4.0）redis-sync-manage：redis-port封装版本（（劣势：不支持密码模式））等工具

刚好线上是redis4.0，有的无的因为各种原因无法使用上面的各种工具

方法二：

参考：【Redis】redis迁移数据到redis-cluster劣势：步骤相对麻烦，档redis rdb文件过大时，迁移时间过长

方法三：本文需要详细说的一种办法（假设集群是三个节点）

1.新的集群（75，74，72）当作slave加入原集群中，等待同步数据2.等待数据同步完毕将其剔除集群3.关闭同步完的三台节点4.编辑nodes.conf文件原nodes.conf应该是这样4ce6e03ba48336569efcc7bbaab6edb538cb1a90 172.23.3.74:6379@16379 slave a7f923aaee624acf672f8704ffb4b27b48a56352 0 1536238939051 19 connectedaa8cf0b64f91c77846862909f9cec0c20ea54c1e 172.23.3.72:6379@16379 slave f2d9ac982c1d621850638a474ed4b8ffc5063d28 0 1536238941056 18 connected2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 172.31.36.20:6379@16379 master - 0 1536238940000 17 connected 0-502 1753-5930 12175-12671f2d9ac982c1d621850638a474ed4b8ffc5063d28 172.31.47.95:6379@16379 master - 0 1536238939000 18 connected 503-1752 5931-6380 10925-11274 12672-163834dd59a9480a669f308a6819a3538e3340ed855cf 172.23.3.75:6379@16379 myself,slave 2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 0 1536238940000 0 connectedc4c7df1e21399984d909761441be80b178ef62a5 172.31.47.95:6380@16380 slave a7f923aaee624acf672f8704ffb4b27b48a56352 0 1536238941000 19 connected9a6307f16ea34354e701c4dbe04ecb5c77021938 172.31.37.24:6380@16380 slave f2d9ac982c1d621850638a474ed4b8ffc5063d28 0 1536238942381 18 connecteda7f923aaee624acf672f8704ffb4b27b48a56352 172.31.37.24:6379@16379 master - 0 1536238942061 19 connected 6381-10924 11275-121745dcc47b5477628895ce98e4c06f8e228a9c4fe5f 172.31.36.20:6380@16380 slave 2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 0 1536238941373 17 connectedvars currentEpoch 19 lastVoteEpoch 0修改nodes.conf为4ce6e03ba48336569efcc7bbaab6edb538cb1a90 172.23.3.74:6379@16379 master - 0 1536238942061 19 connected 6381-10924 11275-12174aa8cf0b64f91c77846862909f9cec0c20ea54c1e 172.23.3.72:6379@16379 master - 0 1536238939000 18 connected 503-1752 5931-6380 10925-11274 12672-163834dd59a9480a669f308a6819a3538e3340ed855cf 172.23.3.75:6379@16379 myself,master - 0 1536238940000 17 connected 0-502 1753-5930 12175-12671vars currentEpoch 19 lastVoteEpoch 0以74为例，修改思路：找到74的主37.24，替换74的slot信息为37.24的slot信息master - 0 1536238942061 19 connected 6381-10924 11275-12174每台机器的master前面分别加上myself信息5.重启75，74，72三台的redis集群并校验可将2-5步写成脚本自动化完成，分钟级就可以复制出一套集群！

补充：还有可能是机房迁移，那样只需要一套集群可用就好了，这个需求更简单！

直接关掉原机房的redis服务，新机房的redis提升为master，修改域名解析就好了，切换过程如下

6379 down机后，6380会被提升为master，6379重启后会变成slave 加入到6380下cluster info变化流程如下：正常情况下[root@ip-172-31-36-20 ~]# /usr/local/redis/bin/redis-cli  -a XSX7sOfvFeLEyXod -h 127.0.0.1 -p 6379 -c cluster nodesf2d9ac982c1d621850638a474ed4b8ffc5063d28 172.31.47.95:6379@16379 master - 0 1536302125000 18 connected 804-1752 5931-6380 10925-11274 12672-1638395c8168dab83f7156c1b0ba34686917972259517 172.31.36.20:6380@16380 slave 2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 0 1536302126000 20 connecteda7f923aaee624acf672f8704ffb4b27b48a56352 172.31.37.24:6379@16379 master - 0 1536302126573 21 connected 787-803 6381-10924 11275-12174596012a79c3699692651fe47bc1a5a982762676a 172.31.47.95:6380@16380 slave f2d9ac982c1d621850638a474ed4b8ffc5063d28 0 1536302124569 21 connecteddd4d8dbe8aee4ef0e31b8908297ca64092302bfd 172.31.37.24:6380@16380 slave a7f923aaee624acf672f8704ffb4b27b48a56352 0 1536302125000 21 connected2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 172.31.36.20:6379@16379 myself,master - 1536291530036 1536291530036 20 connected 0-786 1753-5930 12175-12671down掉master 6379后[root@ip-172-31-36-20 ~]# /usr/local/redis/bin/redis-cli  -a XSX7sOfvFeLEyXod -h 172.31.37.24 -p 6379 -c cluster nodes|grep -i 36a7f923aaee624acf672f8704ffb4b27b48a56352 172.31.37.24:6379@16379 myself,master - 1536291532036 1536291532036 21 connected 787-803 6381-10924 11275-1217495c8168dab83f7156c1b0ba34686917972259517 172.31.36.20:6380@16380 slave 2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 0 1536302197019 20 connectedf2d9ac982c1d621850638a474ed4b8ffc5063d28 172.31.47.95:6379@16379 master - 0 1536302196019 18 connected 804-1752 5931-6380 10925-11274 12672-16383596012a79c3699692651fe47bc1a5a982762676a 172.31.47.95:6380@16380 slave f2d9ac982c1d621850638a474ed4b8ffc5063d28 0 1536302198019 18 connecteddd4d8dbe8aee4ef0e31b8908297ca64092302bfd 172.31.37.24:6380@16380 slave a7f923aaee624acf672f8704ffb4b27b48a56352 0 1536302195000 21 connected2d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 172.31.36.20:6379@16379 master,fail - 1536302172919 1536302169019 20 disconnected 0-786 1753-5930 12175-12671几秒后选举成功，6380提升[root@ip-172-31-36-20 ~]# /usr/local/redis/bin/redis-cli  -a XSX7sOfvFeLEyXod -h 172.31.37.24 -p 6379 -c cluster nodes|grep -i 36.2095c8168dab83f7156c1b0ba34686917972259517 172.31.36.20:6380@16380 master - 0 1536302200019 22 connected 0-786 1753-5930 12175-126712d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 172.31.36.20:6379@16379 master,fail - 1536302172919 1536302169019 20 disconnected重启6379几秒后重新加入集群[root@ip-172-31-36-20 ~]# /usr/local/redis/bin/redis-cli  -a XSX7sOfvFeLEyXod -h 172.31.37.24 -p 6379 -c cluster nodes|grep -i 36.2095c8168dab83f7156c1b0ba34686917972259517 172.31.36.20:6380@16380 master - 0 1536302275023 22 connected 0-786 1753-5930 12175-126712d21ca5d5504b6e8bf283d4a3d0f390ab0258b90 172.31.36.20:6379@16379 slave 95c8168dab83f7156c1b0ba34686917972259517 0 1536302273023 22 connected