2万字好文！手把手教你实现 Docker 部署 Redis 集群

点击上方 "程序员小乐"关注, 星标或置顶一起成长

后台回复“大礼包”有惊喜礼包！

关注订阅号「程序员小乐」，收看更多精彩内容

每日英文

Life is not always what we want it to be. We fight. We cry. And sometimes, we give up. But in our hearts, we know it's still love.

生活有时不尽如人意。我们挣扎、哭泣，有时甚至放弃。但内心始终充满爱.

每日掏心话

人有时需要内心单纯的像个孩子，但有时也要内心强大的像混凝土。

来自：我为什么要写这个 | 责编：乐乐

链接：cnblogs.com/cxbhakim/p/9151720.html

后端架构师(ID:study_tech)第 1067 次推文

往日回顾：任正非在荣耀送别会上的讲话：一旦“离婚”就不要藕断丝连，要做华为全球最强的竞争对手

正文

摘要

接触docker以来，似乎养成了一种习惯，安装什么应用软件都想往docker方向做，今天就想来尝试下使用docker搭建redis集群。

首先，我们需要理论知识：Redis Cluster是Redis的分布式解决方案，它解决了redis单机中心化的问题，分布式数据库——首要解决把整个数据集按照分区规则映射到多个节点的问题。

这边就需要知道分区规则——哈希分区规则。Redis Cluster 采用哈希分区规则中的虚拟槽分区。所有的键根据哈希函数映射到0 ～ 16383，计算公式：slot = CRC16(key)&16383。每一个节点负责维护一部分槽以及槽所映射的键值数据。

一、创建redis docker基础镜像

1.下载redis安装包，使用版本为：4.0.1

[root@etcd1 tmp]# mkdir docker_redis_cluster
[root@etcd1 tmp]# cd docker_redis_cluster/
[root@etcd2 docker_redis_cluster]# wget http://download.redis.io/releases/redis-4.0.1.tar.gz

2.解压编译redis

[root@etcd1 docker_redis_cluster]# tar zxvf redis-4.0.1.tar.gz
[root@etcd1 docker_redis_cluster]# cd redis-4.0.1/
[root@etcd1 redis-4.0.1]# make

3.修改redis配置

[root@etcd3 redis-4.0.1]# vi /tmp/docker_redis_cluster/redis-4.0.1/redis.conf

修改bind ip地址

# ~~~ WARNING ~~~ If the computer running Redis is directly exposed to the
# internet, binding to all the interfaces is dangerous and will expose the
# instance to everybody on the internet. So by default we uncomment the
# following bind directive, that will force Redis to listen only into
# the IPv4 lookback interface address (this means Redis will be able to
# accept connections only from clients running into the same computer it
# is running).
#
# IF YOU ARE SURE YOU WANT YOUR INSTANCE TO LISTEN TO ALL THE INTERFACES
# JUST COMMENT THE FOLLOWING LINE.
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
#bind 127.0.0.1
bind 0.0.0.0

将守护进程yes改成no

# By default Redis does not run as a daemon. Use 'yes' if you need it.
# Note that Redis will write a pid file in /var/run/redis.pid when daemonized.
daemonize no

将密码项注释去掉，添加新密码

# Warning: since Redis is pretty fast an outside user can try up to
# 150k passwords per second against a good box. This means that you should
# use a very strong password otherwise it will be very easy to break.
#
# requirepass foobared

修改为

因为配置了密码，所以，配置中另外一处主从连接也需要配置密码

# If the master is password protected (using the "requirepass" configuration
# directive below) it is possible to tell the slave to authenticate before
# starting the replication synchronization process, otherwise the master will
# refuse the slave request.
#
# masterauth <master-password>

修改为

设置日志路径

# Specify the log file name. Also the empty string can be used to force
# Redis to log on the standard output. Note that if you use standard
# output for logging but daemonize, logs will be sent to /dev/null
logfile "/var/log/redis/redis-server.log"

配置集群相关信息，去掉配置项前面的注释

# Normal Redis instances can't be part of a Redis Cluster; only nodes that are
# started as cluster nodes can. In order to start a Redis instance as a
# cluster node enable the cluster support uncommenting the following:
#
cluster-enabled yes

# Every cluster node has a cluster configuration file. This file is not
# intended to be edited by hand. It is created and updated by Redis nodes.
# Every Redis Cluster node requires a different cluster configuration file.
# Make sure that instances running in the same system do not have
# overlapping cluster configuration file names.
#
cluster-config-file nodes-6379.conf

# Cluster node timeout is the amount of milliseconds a node must be unreachable
# for it to be considered in failure state.
# Most other internal time limits are multiple of the node timeout.
#
cluster-node-timeout 15000

4.镜像制作

[root@etcd3 docker_redis_cluster]# cd /tmp/docker_redis_cluster
[root@etcd3 docker_redis_cluster]# vi Dockerfile
# Redis
# Version 4.0.1

FROM Centos:7<br>
ENV REDIS_HOME /usr/local<br>
ADD redis-4.0.1.tar.gz / # 本地的redis源码包复制到镜像的根路径下，ADD命令会在复制过后自动解包。被复制的对象必须处于Dockerfile同一路径，且ADD后面必须使用相对路径
RUN mkdir -p $REDIS_HOME/redis # 创建安装目录
ADD redis-4.0.1/redis.conf $REDIS_HOME/redis/  # 将一开始编译产生并修改后的配置复制到安装目录

RUN yum -y update  # 更新yum源
RUN yum install -y gcc make # 安装编译需要的工具

WORKDIR /redis-4.0.1
RUN make
RUN mv /redis-4.0.1/src/redis-server  $REDIS_HOME/redis/   # 编译后，容器中只需要可执行文件redis-server

WORKDIR /
RUN rm -rf /redis-4.0.1          # 删除解压文件

RUN yum remove -y gcc make   # 安装编译完成之后，可以删除多余的gcc跟make

VOLUME ["/var/log/redis"]  # 添加数据卷

EXPOSE 6379   # 暴露6379端口，也可以暴露多个端口，这里不需要如此

PS.当前镜像非可执行镜像，所以没有包含ENTRYPOINT和CMD指令

5.构建镜像

# 切换中国源
[root@etcd3 docker_redis_cluster]# vi /etc/docker/daemon.json
{
"registry-mirrors": ["https://registry.docker-cn.com"]
}

# 编译
[root@etcd3 docker_redis_cluster]# docker build -t hakimdstx/cluster-redis .
...

Complete!
---> 546cb1d34f35
Removing intermediate container 6b6556c5f28d
Step 14/15 : VOLUME /var/log/redis
---> Running in 05a6642e4046
---> e7e2fb8676b2
Removing intermediate container 05a6642e4046
Step 15/15 : EXPOSE 6379
---> Running in 5d7abe1709e2
---> 2d1322475f79
Removing intermediate container 5d7abe1709e2
Successfully built 2d1322475f79

镜像制作完成，制作中间可能会报： Public key for glibc-headers-2.17-222.el7.x86_64.rpm is not installed 错误，这时候需要在镜像配置中添加一句命令：

...
RUN rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7
RUN yum -y update # 更新yum源
RUN yum install -y gcc make # 安装编译需要的工具

查看镜像：

[root@etcd3 docker_redis_cluster]# docker images
REPOSITORY                                  TAG                 IMAGE ID            CREATED             SIZE
hakimdstx/cluster-redis                     4.0.1               1fca5a08a4c7        14 seconds ago      435 MB
centos                                      7                   49f7960eb7e4        2 days ago          200 MB

以上，redis 基础镜像就制作完成了

二、制作redis节点镜像

1.基于此前制作的redis基础镜像创建一个redis节点镜像

[root@etcd3 tmp]# mkdir docker_redis_nodes
[root@etcd3 tmp]# cd docker_redis_nodes
[root@etcd3 docker_redis_nodes]# vi Dockerfile
# Redis Node
# Version 4.0.1<br>
FROM hakimdstx/cluster-redis:4.0.1

# MAINTAINER_INFO
MAINTAINER hakim 1194842583@qq.com

ENTRYPOINT ["/usr/local/redis/redis-server", "/usr/local/redis/redis.conf"]

2.构建redis节点镜像

[root@etcd3 docker_redis_nodes]# docker build -t hakimdstx/nodes-redis:4.0.1 .
Sending build context to Docker daemon 2.048 kB
Step 1/3 : FROM hakimdstx/cluster-redis:4.0.1
---> 1fca5a08a4c7
Step 2/3 : MAINTAINER hakim 1194842583@qq.com
---> Running in cc6e07eb2c36
---> 55769d3bfacb
Removing intermediate container cc6e07eb2c36
Step 3/3 : ENTRYPOINT /usr/local/redis/redis-server /usr/local/redis/redis.conf
---> Running in f5dedf88f6f6
---> da64da483559
Removing intermediate container f5dedf88f6f6
Successfully built da64da483559

3.查看镜像

[root@etcd3 docker_redis_nodes]# docker images
REPOSITORY                                  TAG                 IMAGE ID            CREATED             SIZE
hakimdstx/nodes-redis                       4.0.1               da64da483559        51 seconds ago      435 MB
hakimdstx/cluster-redis                     4.0.1               1fca5a08a4c7        9 minutes ago       435 MB
centos                                      7                   49f7960eb7e4        2 days ago          200 MB

三、运行redis集群

1.运行redis容器

[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6379 -p 6379:6379 hakimdstx/nodes-redis:4.0.1
1673a7d859ea83257d5bf14d82ebf717fb31405c185ce96a05f597d8f855aa7d
[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6380 -p 6380:6379 hakimdstx/nodes-redis:4.0.1
df6ebce6f12a6f3620d5a29adcfbfa7024e906c3af48f21fa7e1fa524a361362
[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6381 -p 6381:6379 hakimdstx/nodes-redis:4.0.1
396e174a1d9235228b3c5f0266785a12fb1ea49efc7ac755c9e7590e17aa1a79
[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6382 -p 6382:6379 hakimdstx/nodes-redis:4.0.1
d9a71dd3f969094205ffa7596c4a04255575cdd3acca2d47fe8ef7171a3be528
[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6383 -p 6383:6379 hakimdstx/nodes-redis:4.0.1
73e4f843d8cb28595456e21b04f97d18ce1cdf8dc56d1150844ba258a3781933
[root@etcd3 docker_redis_nodes]# docker run -d --name redis-6384 -p 6384:6379 hakimdstx/nodes-redis:4.0.1
10c62aafa4dac47220daf5bf3cec84406f086d5261599b54ec6c56bb7da97d6d

2.查看容器信息

3.运行 redis 集群容器

a.通过远程连接，查看redis info replication 信息

在公众号程序员小乐后台回复“Java”，获取Java面试题和答案。

[root@etcd2 ~]# redis-cli -h 192.168.10.52 -p 6379
192.168.10.52:6379> info replication
NOAUTH Authentication required.
192.168.10.52:6379> auth 123456
OK
192.168.10.52:6379> info replication
# Replication
role:master
connected_slaves:0
master_replid:2f0a7b50aed699fa50a79f3f7f9751a070c50ee9
master_replid2:0000000000000000000000000000000000000000
master_repl_offset:0
second_repl_offset:-1
repl_backlog_active:0
repl_backlog_size:1048576
repl_backlog_first_byte_offset:0
repl_backlog_histlen:0
192.168.10.52:6379>
# 其余基本信息同上

可以看到，客户连接之后，因为之前设置了密码，所以需要先输入密码认证，否则就无法通过。以上信息，我们知道所有的redis都是master角色 role:master ，这显然不是我们所希望的。

b.在配置之前我们需要查看所有容器当前的IP地址

[root@etcd3 redis]# docker ps
CONTAINER ID        IMAGE                         COMMAND                  CREATED             STATUS              PORTS                    NAMES
10c62aafa4da        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   3 seconds ago       Up 2 seconds        0.0.0.0:6384->6379/tcp   redis-6384
73e4f843d8cb        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   12 seconds ago      Up 10 seconds       0.0.0.0:6383->6379/tcp   redis-6383
d9a71dd3f969        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   20 seconds ago      Up 18 seconds       0.0.0.0:6382->6379/tcp   redis-6382
396e174a1d92        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   3 days ago          Up 3 days           0.0.0.0:6381->6379/tcp   redis-6381
df6ebce6f12a        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   3 days ago          Up 3 days           0.0.0.0:6380->6379/tcp   redis-6380
1673a7d859ea        hakimdstx/nodes-redis:4.0.1   "/usr/local/redis/..."   3 days ago          Up 3 days           0.0.0.0:6379->6379/tcp   redis-6379
[root@etcd3 redis]#
[root@etcd3 redis]# docker inspect 10c62aafa4da 73e4f843d8cb d9a71dd3f969 396e174a1d92 df6ebce6f12a 1673a7d859ea | grep IPA
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.7",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.7",
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.6",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.6",
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.5",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.5",
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.4",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.4",
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.3",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.3",
            "SecondaryIPAddresses": null,
            "IPAddress": "172.17.0.2",
                    "IPAMConfig": null,
                    "IPAddress": "172.17.0.2",

可以知道：redis-6379：172.17.0.2，redis-6380：172.17.0.3，redis-6381：172.17.0.4，redis-6382：172.17.0.5，redis-6383：172.17.0.6，redis-6384：172.17.0.7

c.配置redis

d.ert

4.Redis Cluster 的集群感知操作

//集群(cluster)
CLUSTER INFO 打印集群的信息
CLUSTER NODES 列出集群当前已知的所有节点（node），以及这些节点的相关信息。

//节点(node)
CLUSTER MEET <ip> <port> 将 ip 和 port 所指定的节点添加到集群当中，让它成为集群的一份子。
CLUSTER FORGET <node_id> 从集群中移除 node_id 指定的节点。
CLUSTER REPLICATE <node_id> 将当前节点设置为 node_id 指定的节点的从节点。
CLUSTER SAVECONFIG 将节点的配置文件保存到硬盘里面。

//槽(slot)
CLUSTER ADDSLOTS <slot> [slot ...] 将一个或多个槽（slot）指派（assign）给当前节点。
CLUSTER DELSLOTS <slot> [slot ...] 移除一个或多个槽对当前节点的指派。
CLUSTER FLUSHSLOTS 移除指派给当前节点的所有槽，让当前节点变成一个没有指派任何槽的节点。
CLUSTER SETSLOT <slot> NODE <node_id> 将槽 slot 指派给 node_id 指定的节点，如果槽已经指派给另一个节点，那么先让另一个节点删除该槽>，然后再进行指派。
CLUSTER SETSLOT <slot> MIGRATING <node_id> 将本节点的槽 slot 迁移到 node_id 指定的节点中。
CLUSTER SETSLOT <slot> IMPORTING <node_id> 从 node_id 指定的节点中导入槽 slot 到本节点。
CLUSTER SETSLOT <slot> STABLE 取消对槽 slot 的导入（import）或者迁移（migrate）。

//键 (key)
CLUSTER KEYSLOT <key> 计算键 key 应该被放置在哪个槽上。
CLUSTER COUNTKEYSINSLOT <slot> 返回槽 slot 目前包含的键值对数量。
CLUSTER GETKEYSINSLOT <slot> <count> 返回 count 个 slot 槽中的键。

redis 集群感知：节点握手——是指一批运行在集群模式的节点通过Gossip协议彼此通信，达到感知对方的过程。

192.168.10.52:6379> CLUSTER MEET 172.17.0.3 6379
OK
192.168.10.52:6379> CLUSTER MEET 172.17.0.4 6379
OK
192.168.10.52:6379> CLUSTER MEET 172.17.0.5 6379
OK
192.168.10.52:6379> CLUSTER MEET 172.17.0.6 6379
OK
192.168.10.52:6379> CLUSTER MEET 172.17.0.7 6379
OK
192.168.10.52:6379> CLUSTER NODES
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 master - 0 1528697195600 1 connected
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528697195600 0 connected
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 master - 0 1528697195600 5 connected
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 master - 0 1528697194000 4 connected
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 master - 0 1528697194995 3 connected
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 myself,master - 0 1528697195000 2 connected

当前已经使这六个节点组成集群，但是现在还无法工作，因为集群节点还没有分配槽（slot）。

a.分配槽信息

查看172.17.0.2:6379 的槽个数

192.168.10.52:6379> CLUSTER INFO
cluster_state:fail
cluster_slots_assigned:0 # 被分配槽的个数为0
cluster_slots_ok:0
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6
cluster_size:0
cluster_current_epoch:5
cluster_my_epoch:2
cluster_stats_messages_ping_sent:260418
cluster_stats_messages_pong_sent:260087
cluster_stats_messages_meet_sent:10
cluster_stats_messages_sent:520515
cluster_stats_messages_ping_received:260086
cluster_stats_messages_pong_received:260328
cluster_stats_messages_meet_received:1
cluster_stats_messages_received:520415

上面看到集群状态是失败的，原因是槽位没有分配，而且需要一次性把16384个槽位完全分配了，集群才可用。

b.分配槽位

分配槽位：CLUSTER ADDSLOTS 槽位，一个槽位只能分配一个节点，16384个槽位必须分配完，不同节点不能冲突。

所以通过脚本进行分配 addslots.sh：

#!/bin/bash
# node1 192.168.10.52   172.17.0.2
n=0
for ((i=n;i<=5461;i++))
do
   /usr/local/bin/redis-cli -h 192.168.10.52 -p 6379 -a 123456  CLUSTER ADDSLOTS $i
done

# node2 192.168.10.52    172.17.0.3
n=5462
for ((i=n;i<=10922;i++))
do
   /usr/local/bin/redis-cli -h 192.168.10.52 -p 6380 -a 123456 CLUSTER ADDSLOTS $i
done

# node3 192.168.10.52    172.17.0.4
n=10923
for ((i=n;i<=16383;i++))
do
   /usr/local/bin/redis-cli -h 192.168.10.52 -p 6381 -a 123456 CLUSTER ADDSLOTS $i
done

其中， -a 123456 表示需要输入的密码。

192.168.10.52:6379> CLUSTER INFO
cluster_state:fail　　　　       # 集群状态为失败
cluster_slots_assigned:16101    # 没有完全分配结束
cluster_slots_ok:16101
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6
cluster_size:3
cluster_current_epoch:5
cluster_my_epoch:2
cluster_stats_messages_ping_sent:266756
cluster_stats_messages_pong_sent:266528
cluster_stats_messages_meet_sent:10
cluster_stats_messages_sent:533294
cluster_stats_messages_ping_received:266527
cluster_stats_messages_pong_received:266666
cluster_stats_messages_meet_received:1
cluster_stats_messages_received:533194<br>
192.168.10.52:6379> CLUSTER INFO
cluster_state:ok                   # 集群状态为成功
cluster_slots_assigned:16384       # 已经全部分配完成
cluster_slots_ok:16384
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6
cluster_size:3
cluster_current_epoch:5
cluster_my_epoch:2
cluster_stats_messages_ping_sent:266757
cluster_stats_messages_pong_sent:266531
cluster_stats_messages_meet_sent:10
cluster_stats_messages_sent:533298
cluster_stats_messages_ping_received:266530
cluster_stats_messages_pong_received:266667
cluster_stats_messages_meet_received:1
cluster_stats_messages_received:533198

综上可知，当全部槽位分配完成之后，集群还是可行的，如果我们手欠，移除一个槽位，那么集群就立马那不行了，自己去试试吧 ——CLUSTER DELSLOTS 0 。

5.如何变成高可用性

以上我们已经搭建了一套完整的可运行的redis cluster，但是每个节点都是单点，这样子可能出现，一个节点挂掉，整个集群因为槽位分配不完全而崩溃，因此，我们需要为每个节点配置副本备用节点。

前面我们已经提前创建了6个备用节点，搭建集群花了三个，因此还有剩下三个直接可以用来做备用副本。

192.168.10.52:6379> CLUSTER INFO
cluster_state:ok
cluster_slots_assigned:16384
cluster_slots_ok:16384
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6 # 总共6个节点
cluster_size:3 # 集群为 3 个节点
cluster_current_epoch:5
cluster_my_epoch:2
cluster_stats_messages_ping_sent:270127
cluster_stats_messages_pong_sent:269893
cluster_stats_messages_meet_sent:10
cluster_stats_messages_sent:540030
cluster_stats_messages_ping_received:269892
cluster_stats_messages_pong_received:270037
cluster_stats_messages_meet_received:1
cluster_stats_messages_received:539930

查看所有节点的id

192.168.10.52:6379> CLUSTER NODES
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 master - 0 1528704114535 1 connected 5462-10922
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528704114000 0 connected 10923-16383
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 master - 0 1528704114023 5 connected
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 master - 0 1528704115544 4 connected
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 master - 0 1528704114836 3 connected
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 myself,master - 0 1528704115000 2 connected 0-5461

编写脚本，添加副本节点

[root@etcd2 tmp]# vi addSlaveNodes.sh
#!/bin/bash

/usr/local/bin/redis-cli -h 192.168.10.52 -p 6382 -a 123456 CLUSTER REPLICATE 760e4d0039c5ac13d04aa4791c9e6dc28544d7c7

/usr/local/bin/redis-cli -h 192.168.10.52 -p 6383 -a 123456 CLUSTER REPLICATE 54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c

/usr/local/bin/redis-cli -h 192.168.10.52 -p 6384 -a 123456 CLUSTER REPLICATE f45f9109f2297a83b1ac36f9e1db5e70bbc174ab

注意：

1、作为备用的节点，必须是未分配槽位的，否者会操作失败 (error) ERR To set a master the node must be empty and without assigned slots 。

2、需要从需要添加的节点上面执行操作，CLUSTER REPLICATE [node_id] ，使当前节点成为 node_id 的副本节点。

3、添加从节点（集群复制），复制的原理和单机的Redis复制原理一样，区别是：集群下的从节点也需要运行在cluster模式下，要先添加到集群里面，再做复制。

查看所有节点信息：

192.168.10.52:6379> CLUSTER NODES
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 master - 0 1528705604149 1 connected 5462-10922
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528705603545 0 connected 10923-16383
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 slave f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 0 1528705603144 5 connected
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 slave 54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 0 1528705603000 4 connected
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 slave 760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 0 1528705603000 3 connected
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 myself,master - 0 1528705602000 2 connected 0-5461

可以看到我们现在实现了三主三从的一个高可用集群。

6.高可用测试——故障转移

查看当前运行状态：

在公众号顶级架构师后台回复“offer”，获取算法面试题和答案。

以上，运行正常

尝试关闭一个master，选择端口为6380的容器，停掉之后：

192.168.10.52:6379> CLUSTER NODES
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 master,fail - 1528706408935 1528706408000 1 connected 5462-10922
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528706463000 0 connected 10923-16383
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 slave f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 0 1528706462980 5 connected
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 slave 54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 0 1528706463000 4 connected
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 slave 760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 0 1528706463985 3 connected
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 myself,master - 0 1528706462000 2 connected 0-5461
192.168.10.52:6379>
192.168.10.52:6379> CLUSTER INFO
cluster_state:fail
cluster_slots_assigned:16384
cluster_slots_ok:10923
cluster_slots_pfail:0
cluster_slots_fail:5461
cluster_known_nodes:6
cluster_size:3
cluster_current_epoch:5
cluster_my_epoch:2
cluster_stats_messages_ping_sent:275112
cluster_stats_messages_pong_sent:274819
cluster_stats_messages_meet_sent:10
cluster_stats_messages_fail_sent:5
cluster_stats_messages_sent:549946
cluster_stats_messages_ping_received:274818
cluster_stats_messages_pong_received:275004
cluster_stats_messages_meet_received:1
cluster_stats_messages_fail_received:1
cluster_stats_messages_received:549824

以上，发现整个集群都失败了，从节点没有自动升级为主节点，怎么回事？？

重启停掉的容器，经排查日志信息 [root@df6ebce6f12a /]# tail -f /var/log/redis/redis-server.log ：

1:S 11 Jun 09:57:46.712 # Cluster state changed: ok
1:S 11 Jun 09:57:46.718 * (Non critical) Master does not understand REPLCONF listening-port: -NOAUTH Authentication required.
1:S 11 Jun 09:57:46.718 * (Non critical) Master does not understand REPLCONF capa: -NOAUTH Authentication required.
1:S 11 Jun 09:57:46.719 * Partial resynchronization not possible (no cached master)
1:S 11 Jun 09:57:46.719 # Unexpected reply to PSYNC from master: -NOAUTH Authentication required.
1:S 11 Jun 09:57:46.719 * Retrying with SYNC...
1:S 11 Jun 09:57:46.719 # MASTER aborted replication with an error: NOAUTH Authentication required.
1:S 11 Jun 09:57:46.782 * Connecting to MASTER 172.17.0.6:6379
1:S 11 Jun 09:57:46.782 * MASTER <-> SLAVE sync started
1:S 11 Jun 09:57:46.782 * Non blocking connect for SYNC fired the event.

可以看到，主从之间访问需要auth，之前忘记了配置 redis.conf 中的 # masterauth <master-password> ，所以导致主从之间无法通讯。修改配置之后，自动故障转移正常。

有时候需要实施人工故障转移

登录6380端口的从节点：6383，执行 CLUSTER FAILOVER 命令：

192.168.10.52:6383> CLUSTER FAILOVER
(error) ERR Master is down or failed, please use CLUSTER FAILOVER FORCE

发现因为master已经down了，所以我们需要执行强制转移

192.168.10.52:6383> CLUSTER FAILOVER FORCE
OK

查看当前 cluster node 情况：

192.168.10.52:6383> CLUSTER NODES
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 slave 760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 0 1528707535332 3 connected
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 slave f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 0 1528707534829 5 connected
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528707534527 0 connected 10923-16383
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 myself,master - 0 1528707535000 6 connected 5462-10922
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 master - 0 1528707535834 2 connected 0-5461
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 master,fail - 1528707472833 1528707472000 1 connected

从节点已经升级为master节点。这时候，我们尝试重启了，6380节点的redis（其实是重新启动停掉的容器）：

192.168.10.52:6383> CLUSTER NODES
0bbdc4176884ef0e3bb9b2e7d03d91b0e7e11f44 172.17.0.5:6379@16379 slave 760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 0 1528707556044 3 connected
ae86224a3bc29c4854719c83979cb7506f37787a 172.17.0.7:6379@16379 slave f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 0 1528707555000 5 connected
f45f9109f2297a83b1ac36f9e1db5e70bbc174ab 172.17.0.4:6379@16379 master - 0 1528707556000 0 connected 10923-16383
98aebcfe42d8aaa8a3375e4a16707107dc9da683 172.17.0.6:6379@16379 myself,master - 0 1528707556000 6 connected 5462-10922
760e4d0039c5ac13d04aa4791c9e6dc28544d7c7 172.17.0.2:6379@16379 master - 0 1528707556000 2 connected 0-5461
54cb5c2eb8e5f5aed2d2f7843f75a9284ef6785c 172.17.0.3:6379@16379 slave 98aebcfe42d8aaa8a3375e4a16707107dc9da683 0 1528707556547 6 connected

我们发现，6380节点反而变成了 6383节点的从节点。

现在集群应该是完整的了，所以，集群状态应该已经恢复了，我们查看下：

192.168.10.52:6383> CLUSTER INFO
cluster_state:ok
cluster_slots_assigned:16384
cluster_slots_ok:16384
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:6
cluster_size:3
cluster_current_epoch:6
cluster_my_epoch:6
cluster_stats_messages_ping_sent:19419
cluster_stats_messages_pong_sent:19443
cluster_stats_messages_meet_sent:1
cluster_stats_messages_auth-req_sent:5
cluster_stats_messages_update_sent:1
cluster_stats_messages_sent:38869
cluster_stats_messages_ping_received:19433
cluster_stats_messages_pong_received:19187
cluster_stats_messages_meet_received:5
cluster_stats_messages_fail_received:4
cluster_stats_messages_auth-ack_received:2
cluster_stats_messages_received:38631

OK，没有问题。

7.集群访问

客户端在初始化的时候只需要知道一个节点的地址即可，客户端会先尝试向这个节点执行命令，比如 get key ，如果key所在的slot刚好在该节点上，则能够直接执行成功。如果slot不在该节点，则节点会返回MOVED错误，同时把该slot对应的节点告诉客户端，客户端可以去该节点执行命令

192.168.10.52:6383> get hello
(error) MOVED 866 172.17.0.2:6379

192.168.10.52:6379> set number 20004
(error) MOVED 7743 172.17.0.3:6379

另外，redis集群版只使用db0，select命令虽然能够支持select 0。其他的db都会返回错误。

192.168.10.52:6383> select 0
OK
192.168.10.52:6383> select 1
(error) ERR SELECT is not allowed in cluster mode

报错问题

docker redis集群连接报错的问题，具体报错如下：

node节点没有全部添加进去，添加之后，依然有上述问题。想到是跨主机访问，应该是路由寻址不了导致的。当初写上述教程的时候，docker是以默认的网络模式bridge模式运行的，毕竟当初是以学习整理文档为主，主要是单机访问。

但是，实际应用化场景中，多是公网跨主机访问，问题明朗了，想着集群这东西最好还是设置成共享主机公网ip比较好，于是解决如下：

1.在docker运行时，执行网络模式为：host。
2.端口冲突解决，毕竟host模式下，容器会占用宿主机的端口，于是，我们就从配置下手，在宿主机上生成配置redis-60001.conf,redis-60002.conf,redis-60003.conf…，有多少端口建多少个文件，最终运行一个容器，挂载一个配置到容器中用于覆盖主机中的配置。

最终的运行方式如下：

docker run -d --name redis-6380 --net host -v /tmp/redis.conf:/usr/local/redis/redis.conf hakimdstx/nodes-redis:4.0.1

至此，网络问题得到解决。