部署redis集群方式的选择

• Statefulset
• Service&depolyment

对于redis,mysql这种有状态的服务,我们使用statefulset方式为首选.我们这边主要就是介绍statefulset这种方式

ps:
statefulset 的设计原理模型:拓扑状态.应用的多个实例之间不是完全对等的关系,这个应用实例的启动必须按照某些顺序启动,比如应用的主节点 A 要先于从节点 B 启动。而如果你把 A 和 B 两个Pod删除掉,他们再次被创建出来是也必须严格按照这个顺序才行,并且,新创建出来的Pod,必须和原来的Pod的网络标识一样,这样原先的访问者才能使用同样的方法,访问到这个新的Pod存储状态:应用的多个实例分别绑定了不同的存储数据.对于这些应用实例来说,Pod A第一次读取到的数据,和隔了十分钟之后再次读取到的数据,应该是同一份,哪怕在此期间Pod A被重新创建过.一个数据库应用的多个存储实例

部署

安装NFS(共享存储)

因为k8s上pod是飘忽不定的,所以我们肯定需要用一个共享存储来提供存储,这样不管pod漂移到哪个节点都能访问这个共享的数据卷.我这个地方先使用NFS来做共享存储,后期可以 选择别的替换

yum -y install nfs-utils rpcbind
vim /etc/exports
/usr/local/kubernetes/redis/pv1 0.0.0.0/0(rw,all_squash)
/usr/local/kubernetes/redis/pv2 0.0.0.0/0(rw,all_squash)
/usr/local/kubernetes/redis/pv3 0.0.0.0/0(rw,all_squash)
/usr/local/kubernetes/redis/pv4 0.0.0.0/0(rw,all_squash)
/usr/local/kubernetes/redis/pv5 0.0.0.0/0(rw,all_squash) /usr/local/kubernetes/redis/pv6 0.0.0.0/0(rw,all_squash) mkdir -p /usr/local/kubernetes/redis/pv{1..6} chmod 777 /usr/local/kubernetes/redis/pv{1..6}

后期我们可以写成域名 通配符

启动服务
systemctl enable nfs
systemctl enable rpcbind
systemctl start nfs
systemctl start rpcbind

创建pv

创建6个pv 一会供pvc挂载使用

vim pv.yaml
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv1
spec:capacity:storage: 200M      #磁盘大小200MaccessModes:- ReadWriteMany    #多客户可读写nfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv1"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-vp2
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv2"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv3
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv3"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv4
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv4"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv5
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv5"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv6
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: NFS服务器地址path: "/usr/local/kubernetes/redis/pv6"

字段说明:
apiversion: api版本
kind: 这个yaml是生成pv的
metadata: 元数据
spec.capacity: 进行资源限制的
spec.accessmodes: 访问模式(读写模式)
spec.nfs: 这个pv卷名是通过nfs提供的

创建pv

kubectl create -f pv.yaml
kubectl get pv   #查看创建的pv

创建configmap,用来存放redis的配置文件

因为redis的配置文件里面可能会改变,所以我们使用configmap这种方式给配置文件弄出来,我们后期改的时候就不需要没改次配置文件就从新生成一个docker images包了

appendonly yes                      #开启Redis的AOF持久化
cluster-enabled yes                 #集群模式打开
cluster-config-file /var/lib/redis/nodes.conf  #下面说明
cluster-node-timeout 5000           #节点超时时间
dir /var/lib/redis                  #AOF持久化文件存在的位置 port 6379 #开启的端口

cluster-conf-file: 选项设定了保存节点配置文件的路径,如果这个配置文件不存在,每个节点在启动的时候都为他自身指定了一个新的ID存档到这个文件中,实例会一直使用同一个ID,在集群中保持一个独一无二的(Unique)名字.每个节点都是用ID而不是IP或者端口号来记录其他节点,因为在k8s中,IP地址是不固定的,而这个独一无二的标识符(Identifier)则会在节点的整个生命周期中一直保持不变,我们这个文件里面存放的是节点ID

创建名为redis-conf的Configmap：

kubectl create configmap redis-conf --from-file=redis.conf

查看:

[root@rke ~]# kubectl get cm
NAME         DATA   AGE
redis-conf   1      22h
[root@rke ~]# kubectl describe cm redis-conf
Name:         redis-conf
Namespace:    default
Labels:       <none>
Annotations:  <none>Data
====
redis.conf:
----
appendonly yes
cluster-enabled yes
cluster-config-file /var/lib/redis/nodes.conf
cluster-node-timeout 5000
dir /var/lib/redis
port 6379Events:  <none>

创建headless service

Headless service是StatefulSet实现稳定网络标识的基础，我们需要提前创建。准备文件headless-service.yml如下：

apiVersion: v1
kind: Service
metadata:name: redis-servicelabels:app: redis
spec:ports:- name: redis-portport: 6379clusterIP: Noneselector:app: redisappCluster: redis-cluster

创建：

kubectl create -f headless-service.yml

查看：

[root@k8s-node1 redis]# kubectl get svc redis-service
NAME            TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
redis-service   ClusterIP   None         <none>        6379/TCP   53s

可以看到，服务名称为redis-service，其CLUSTER-IP为None，表示这是一个“无头”服务。

创建redis集群节点

这是本文的核心内容,创建redis.yaml文件

[root@rke ~]# cat /home/docker/redis/redis.yml
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:name: redis-app
spec:serviceName: "redis-service"replicas: 6template:metadata:labels:app: redisappCluster: redis-clusterspec:terminationGracePeriodSeconds: 20affinity:podAntiAffinity:preferredDuringSchedulingIgnoredDuringExecution:- weight: 100podAffinityTerm:labelSelector:matchExpressions:- key: appoperator: Invalues:- redistopologyKey: kubernetes.io/hostnamecontainers:- name: redisimage: "redis"command:- "redis-server"                  #redis启动命令args:- "/etc/redis/redis.conf"         #redis-server后面跟的参数,换行代表空格- "--protected-mode"              #允许外网访问- "no"# command: redis-server /etc/redis/redis.conf --protected-mode noresources:                          #资源requests:                         #请求的资源cpu: "100m"                     #m代表千分之,相当于0.1 个cpu资源memory: "100Mi"                 #内存100m大小ports:- name: rediscontainerPort: 6379protocol: "TCP"- name: clustercontainerPort: 16379protocol: "TCP"volumeMounts:- name: "redis-conf"              #挂载configmap生成的文件mountPath: "/etc/redis"         #挂载到哪个路径下- name: "redis-data"              #挂载持久卷的路径mountPath: "/var/lib/redis"volumes:- name: "redis-conf"                  #引用configMap卷configMap:name: "redis-conf"items:- key: "redis.conf"             #创建configMap指定的名称path: "redis.conf"            #里面的那个文件--from-file参数后面的文件volumeClaimTemplates:                     #进行pvc持久卷声明,- metadata:name: redis-dataspec:accessModes:- ReadWriteManyresources:requests:storage: 200M

PodAntiAffinity:表示反亲和性，其决定了某个pod不可以和哪些Pod部署在同一拓扑域，可以用于将一个服务的POD分散在不同的主机或者拓扑域中，提高服务本身的稳定性。
matchExpressions:规定了Redis Pod要尽量不要调度到包含app为redis的Node上，也即是说已经存在Redis的Node上尽量不要再分配Redis Pod了.
另外，根据StatefulSet的规则，我们生成的Redis的6个Pod的hostname会被依次命名为$(statefulset名称)-$(序号)，如下图所示：

[root@rke ~]# kubectl get pods -o wide
NAME                                     READY   STATUS    RESTARTS   AGE   IP           NODE            NOMINATED NODE
redis-app-0                              1/1     Running   0          40m   10.42.2.17   192.168.1.21    <none>
redis-app-1                              1/1     Running   0          40m   10.42.0.15   192.168.1.114   <none>
redis-app-2                              1/1     Running   0          40m   10.42.1.13   192.168.1.20    <none>
redis-app-3                              1/1     Running   0          40m   10.42.2.18   192.168.1.21    <none>
redis-app-4                              1/1     Running   0          40m   10.42.0.16   192.168.1.114   <none>
redis-app-5                              1/1     Running   0          40m   10.42.1.14   192.168.1.20    <none>

如上，可以看到这些Pods在部署时是以{0..N-1}的顺序依次创建的。注意，直到redis-app-0状态启动后达到Running状态之后，redis-app-1 才开始启动。
同时，每个Pod都会得到集群内的一个DNS域名，格式为$(podname).$(service name).$(namespace).svc.cluster.local，也即是：

redis-app-0.redis-service.default.svc.cluster.local
redis-app-1.redis-service.default.svc.cluster.local
...以此类推...

在K8S集群内部，这些Pod就可以利用该域名互相通信。我们可以使用busybox镜像的nslookup检验这些域名:

[root@k8s-node1 ~]# kubectl run -i --tty --image busybox dns-test --restart=Never --rm /bin/sh
If you don't see a command prompt, try pressing enter.
/ # nslookup redis-app-1.redis-service.default.svc.cluster.local
Server:     10.43.0.10
Address:    10.43.0.10:53Name:   redis-app-1.redis-service.default.svc.cluster.local
Address: 10.42.0.15*** Can't find redis-app-1.redis-service.default.svc.cluster.local: No answer/ # nslookup redis-app-0.redis-service.default.svc.cluster.local
Server:     10.43.0.10
Address:    10.43.0.10:53Name:   redis-app-0.redis-service.default.svc.cluster.local
Address: 10.42.2.17

可以看到， redis-app-0的IP为10.42.2.17。当然，若Redis Pod迁移或是重启（我们可以手动删除掉一个Redis Pod来测试），则IP是会改变的，但Pod的域名、SRV records、A record都不会改变。
另外可以发现，我们之前创建的pv都被成功绑定了：

[root@k8s-node1 ~]# kubectl get pv
NAME      CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS    CLAIM                            STORAGECLASS   REASON    AGE
nfs-pv1   200M       RWX            Retain           Bound     default/redis-data-redis-app-2                            1h
nfs-pv2   200M       RWX            Retain           Bound     default/redis-data-redis-app-3                            1h
nfs-pv3   200M       RWX            Retain           Bound     default/redis-data-redis-app-4                            1h
nfs-pv4   200M       RWX            Retain           Bound     default/redis-data-redis-app-5                            1h
nfs-pv5   200M       RWX            Retain           Bound     default/redis-data-redis-app-0                            1h
nfs-pv6   200M       RWX            Retain           Bound     default/redis-data-redis-app-1                            1h

初始化redis集群

创建好6个Redis Pod后，我们还需要利用常用的Redis-tribe工具进行集群的初始化。

创建centos容器

由于Redis集群必须在所有节点启动后才能进行初始化，而如果将初始化逻辑写入Statefulset中，则是一件非常复杂而且低效的行为。这里，本人不得不称赞一下原项目作者的思路，值得学习。也就是说，我们可以在K8S上创建一个额外的容器，专门用于进行K8S集群内部某些服务的管理控制。
这里，我们专门启动一个Ubuntu的容器，可以在该容器中安装Redis-tribe，进而初始化Redis集群，执行：

kubectl run -i --tty centos --image=centos --restart=Never /bin/bash

成功后，我们可以进入centos容器中，原项目要求执行如下命令安装基本的软件环境:

cat >> /etc/yum.repo.d/epel.repo<<'EOF'
[epel]
name=Extra Packages for Enterprise Linux 7 - $basearch
baseurl=https://mirrors.tuna.tsinghua.edu.cn/epel/7/$basearch
#mirrorlist=https://mirrors.fedoraproject.org/metalink?repo=epel-7&arch=$basearch failovermethod=priority enabled=1 gpgcheck=0 gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-EPEL-7 EOF

初始化redis集群
首先，我们需要安装redis-trib(redis集群命令行工具)：

yum -y install redis-trib.noarch bind-utils-9.9.4-72.el7.x86_64

然后创建一主一从的集群节点信息：

redis-trib create --replicas 1 \
`dig +short redis-app-0.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-1.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-2.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-3.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-4.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-5.redis-service.default.svc.cluster.local`:6379#create: 创建一个新的集群
#--replicas 1 : 创建的集群中每个主节点分配一个从节点,达到3主3从
#后面跟的就是redis实例所在的位置

如上，命令dig +short redis-app-0.redis-service.default.svc.cluster.local用于将Pod的域名转化为IP，这是因为redis-trib不支持域名来创建集群。
执行完成后redis-trib会打印一份预配置文件给你查看,如果没问题输入yes,redis-trib就会把这份配置文件应用到集群中

>>> Creating cluster
>>> Performing hash slots allocation on 6 nodes...
Using 3 masters:
10.42.2.17:6379
10.42.0.15:6379
10.42.1.13:6379
Adding replica 10.42.2.18:6379 to 10.42.2.17:6379
Adding replica 10.42.0.16:6379 to 10.42.0.15:6379
Adding replica 10.42.1.14:6379 to 10.42.1.13:6379
M: 4676f8913cdcd1e256db432531c80591ae6c5fc3 10.42.2.17:6379slots:0-5460 (5461 slots) master
M: 505f3e126882c0c5115885e54f9b361bc7e74b97 10.42.0.15:6379slots:5461-10922 (5462 slots) master
M: 589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 10.42.1.13:6379slots:10923-16383 (5461 slots) master
S: 366abbba45d3200329a5c6305fbcec9e29b50c80 10.42.2.18:6379replicates 4676f8913cdcd1e256db432531c80591ae6c5fc3
S: cee3a27cc27635da54d94f16f6375cd4acfe6c30 10.42.0.16:6379 replicates 505f3e126882c0c5115885e54f9b361bc7e74b97 S: e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 10.42.1.14:6379 replicates 589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f Can I set the above configuration? (type 'yes' to accept):

输入yes后开始创建集群

>>> 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 10.42.2.17:6379)
M: 4676f8913cdcd1e256db432531c80591ae6c5fc3 10.42.2.17:6379slots:0-5460 (5461 slots) master1 additional replica(s)
M: 589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 10.42.1.13:6379@16379slots:10923-16383 (5461 slots) master1 additional replica(s)
S: e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 10.42.1.14:6379@16379slots: (0 slots) slavereplicates 589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f S: 366abbba45d3200329a5c6305fbcec9e29b50c80 10.42.2.18:6379@16379 slots: (0 slots) slave replicates 4676f8913cdcd1e256db432531c80591ae6c5fc3 M: 505f3e126882c0c5115885e54f9b361bc7e74b97 10.42.0.15:6379@16379 slots:5461-10922 (5462 slots) master 1 additional replica(s) S: cee3a27cc27635da54d94f16f6375cd4acfe6c30 10.42.0.16:6379@16379 slots: (0 slots) slave replicates 505f3e126882c0c5115885e54f9b361bc7e74b97 [OK] All nodes agree about slots configuration. >>> Check for open slots... >>> Check slots coverage... [OK] All 16384 slots covered.

最后一句表示集群中的16384个槽都有至少一个主节点在处理， 集群运作正常.

至此，我们的Redis集群就真正创建完毕了，连到任意一个Redis Pod中检验一下：

root@k8s-node1 ~]# kubectl exec -it redis-app-2 /bin/bash
root@redis-app-2:/data# /usr/local/bin/redis-cli -c
127.0.0.1:6379> 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
cluster_size:3
cluster_current_epoch:6
cluster_my_epoch:1
cluster_stats_messages_ping_sent:186
cluster_stats_messages_pong_sent:199
cluster_stats_messages_sent:385
cluster_stats_messages_ping_received:194
cluster_stats_messages_pong_received:186
cluster_stats_messages_meet_received:5
cluster_stats_messages_received:385127.0.0.1:6379> cluster nodes
589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 10.42.1.13:6379@16379 master - 0 1550555011000 3 connected 10923-16383
e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 10.42.1.14:6379@16379 slave 589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 0 1550555011512 6 connected 366abbba45d3200329a5c6305fbcec9e29b50c80 10.42.2.18:6379@16379 slave 4676f8913cdcd1e256db432531c80591ae6c5fc3 0 1550555010507 4 connected 505f3e126882c0c5115885e54f9b361bc7e74b97 10.42.0.15:6379@16379 master - 0 1550555011000 2 connected 5461-10922 cee3a27cc27635da54d94f16f6375cd4acfe6c30 10.42.0.16:6379@16379 slave 505f3e126882c0c5115885e54f9b361bc7e74b97 0 1550555011713 5 connected 4676f8913cdcd1e256db432531c80591ae6c5fc3 10.42.2.17:6379@16379 myself,master - 0 1550555010000 1 connected 0-5460

另外，还可以在NFS上查看Redis挂载的数据：

[root@rke ~]# tree /usr/local/kubernetes/redis/
/usr/local/kubernetes/redis/
├── pv1
│   ├── appendonly.aof
│   ├── dump.rdb
│   └── nodes.conf
├── pv2
│   ├── appendonly.aof
│   ├── dump.rdb
│   └── nodes.conf
├── pv3
│   ├── appendonly.aof
│   ├── dump.rdb
│   └── nodes.conf
├── pv4
│   ├── appendonly.aof
│   ├── dump.rdb
│   └── nodes.conf
├── pv5
│   ├── appendonly.aof
│   ├── dump.rdb
│   └── nodes.conf
└── pv6├── appendonly.aof├── dump.rdb└── nodes.conf6 directories, 18 files

创建用于访问service

前面我们创建了用于实现statefulset的headless service,但该service没有cluster IP,因此不能用于外界访问.所以我们还需要创建一个service,专用于为Redis集群提供访问和负载均衡:

piVersion: v1
kind: Service
metadata:name: redis-access-servicelabels:app: redis
spec:ports:- name: redis-portprotocol: "TCP"port: 6379targetPort: 6379selector:app: redisappCluster: redis-cluster

如上，该Service名称为 redis-access-service，在K8S集群中暴露6379端口，并且会对labels name为app: redis或appCluster: redis-cluster的pod进行负载均衡。
创建后查看：

[root@rke ~]# kubectl get svc redis-access-service -o wide
NAME                   TYPE        CLUSTER-IP    EXTERNAL-IP   PORT(S)    AGE   SELECTOR
redis-access-service   ClusterIP   10.43.40.62   <none>        6379/TCP   47m   app=redis,appCluster=redis-cluster

如上,在k8s集群中,所有应用都可以通过10.43.40.62:6379来访问redis集群,当然，为了方便测试，我们也可以为Service添加一个NodePort映射到物理机上，待测试。

测试主从切换

在K8S上搭建完好Redis集群后，我们最关心的就是其原有的高可用机制是否正常。这里，我们可以任意挑选一个Master的Pod来测试集群的主从切换机制，如redis-app-2：

[root@rke ~]# kubectl get pods redis-app-2 -o wide
NAME          READY   STATUS    RESTARTS   AGE   IP           NODE           NOMINATED NODE
redis-app-2   1/1     Running   0          1h    10.42.1.13   192.168.1.20   <none>

进入redis-app-2查看：

[root@rke ~]# kubectl exec -it redis-app-2 /bin/bash
root@redis-app-2:/data# redis-cli
127.0.0.1:6379> role
1) "master"
2) (integer) 9478
3) 1) 1) "10.42.1.14" 2) "6379" 3) "9478"

如上可以看到，其为master，slave为10.42.1.14即redis-app-5。

接着，我们手动删除redis-app-2：

[root@rke ~]# kubectl delete pods redis-app-2
pod "redis-app-2" deleted
[root@rke ~]# kubectl get pods redis-app-2 -o wide
NAME          READY   STATUS    RESTARTS   AGE   IP           NODE           NOMINATED NODE
redis-app-2   1/1     Running   0          19s   10.42.1.15   192.168.1.20   <none>

如上，IP改变为10.42.1.15。我们再进入redis-app-2内部查看：

[root@rke ~]# kubectl exec -it redis-app-2 /bin/bash
root@redis-app-2:/data# redis-cli
127.0.0.1:6379> ROLE
1) "slave"
2) "10.42.1.14"
3) (integer) 6379 4) "connected" 5) (integer) 9688

如上，redis-app-2变成了slave，从属于它之前的从节点10.42.1.14即redis-app-5。

redis动态扩容

我们现在这个集群中有6个节点三主三从,我现在添加两个pod节点,达到4主4从

添加nfs共享目录

cat >> /etc/exports <<'EOF'
/usr/local/kubernetes/redis/pv7 192.168.0.0/16(rw,all_squash)
/usr/local/kubernetes/redis/pv8 192.168.0.0/16(rw,all_squash)
EOF
systemctl restart nfs rpcbind
[root@rke ~]# mkdir /usr/local/kubernetes/redis/pv{7..8}
[root@rke ~]# chmod 777 /usr/local/kubernetes/redis/*

更新pv的yml文件,也可以自己在重新创建一个,这边选择自己新建

[root@rke redis]# cat pv_add.yml
---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv7
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: 192.168.1.253path: "/usr/local/kubernetes/redis/pv7"---
apiVersion: v1
kind: PersistentVolume
metadata:name: nfs-pv8
spec:capacity:storage: 200MaccessModes:- ReadWriteManynfs:server: 192.168.1.253path: "/usr/local/kubernetes/redis/pv8"

创建查看pv:

[root@rke redis]# kubectl create -f pv_add.yml
persistentvolume/nfs-pv7 created
persistentvolume/nfs-pv8 created
[root@rke redis]# kubectl get pv
NAME          CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS      CLAIM                            STORAGECLASS   REASON   AGE
nfs-pv1       200M       RWX            Retain           Bound       default/redis-data-redis-app-1                           2h
nfs-pv2       200M       RWX            Retain           Bound       default/redis-data-redis-app-2                           2h
nfs-pv3       200M       RWX            Retain           Bound       default/redis-data-redis-app-4                           2h
nfs-pv4       200M       RWX            Retain           Bound       default/redis-data-redis-app-5                           2h
nfs-pv5       200M       RWX            Retain           Bound       default/redis-data-redis-app-0                           2h
nfs-pv6       200M       RWX            Retain           Bound       default/redis-data-redis-app-3                           2h
nfs-pv7       200M       RWX            Retain           Available                                                            7s
nfs-pv8       200M       RWX            Retain           Available                                                            7s

添加redis节点

更改redis的yml文件里面的replicas:字段,把这个字段改为8,然后升级运行

[root@rke redis]# kubectl apply -f redis.yml
Warning: kubectl apply should be used on resource created by either kubectl create --save-config or kubectl apply
statefulset.apps/redis-app configured[root@rke redis]# kubectl get  pods
NAME                                     READY   STATUS    RESTARTS   AGE
redis-app-0                              1/1     Running   0          2h
redis-app-1                              1/1     Running   0          2h
redis-app-2                              1/1     Running   0          19m
redis-app-3                              1/1     Running   0          2h
redis-app-4                              1/1     Running   0          2h
redis-app-5                              1/1     Running   0          2h
redis-app-6                              1/1     Running   0          57s
redis-app-7                              1/1     Running   0          30s

添加集群节点

[root@rke redis]#kubectl exec -it centos /bin/bash
[root@centos /]# redis-trib add-node \
`dig +short redis-app-6.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-0.redis-service.default.svc.cluster.local`:6379[root@centos /]# redis-trib add-node \
`dig +short redis-app-7.redis-service.default.svc.cluster.local`:6379 \
`dig +short redis-app-0.redis-service.default.svc.cluster.local`:6379

add-node后面跟的是新节点的信息,后面是以前集群中的任意 一个节点

查看添加redis节点是否正常

[root@rke redis]# kubectl exec -it redis-app-0 bash
root@redis-app-0:/data# redis-cli
127.0.0.1:6379> cluster nodes
589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 10.42.1.15:6379@16379 slave e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 0 1550564776000 7 connected
e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 10.42.1.14:6379@16379 master - 0 1550564776000 7 connected 10923-16383 366abbba45d3200329a5c6305fbcec9e29b50c80 10.42.2.18:6379@16379 slave 4676f8913cdcd1e256db432531c80591ae6c5fc3 0 1550564777051 4 connected 505f3e126882c0c5115885e54f9b361bc7e74b97 10.42.0.15:6379@16379 master - 0 1550564776851 2 connected 5461-10922 cee3a27cc27635da54d94f16f6375cd4acfe6c30 10.42.0.16:6379@16379 slave 505f3e126882c0c5115885e54f9b361bc7e74b97 0 1550564775000 5 connected e4697a7ba460ae2979692116b95fbe1f2c8be018 10.42.0.20:6379@16379 master - 0 1550564776549 0 connected 246c79682e6cc78b4c2c28d0e7166baf47ecb265 10.42.2.23:6379@16379 master - 0 1550564776548 8 connected 4676f8913cdcd1e256db432531c80591ae6c5fc3 10.42.2.17:6379@16379 myself,master - 0 1550564775000 1 connected 0-5460

重新分配哈希槽

redis-trib.rb reshard `dig +short redis-app-0.redis-service.default.svc.cluster.local`:6379
## 输入要移动的哈希槽
## 移动到哪个新的master节点(ID)
## all 是从所有master节点上移动

查看对应的节点信息

127.0.0.1:6379> cluster nodes
589b4f4f908a04f56d2ab9cd6fd0fd25ea14bb8f 10.42.1.15:6379@16379 slave e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 0 1550566162000 7 connected
e9f1f704ff7c8f060d6b39e23be9cd8e55cb2e46 10.42.1.14:6379@16379 master - 0 1550566162909 7 connected 11377-16383
366abbba45d3200329a5c6305fbcec9e29b50c80 10.42.2.18:6379@16379 slave 4676f8913cdcd1e256db432531c80591ae6c5fc3 0 1550566161600 4 connected
505f3e126882c0c5115885e54f9b361bc7e74b97 10.42.0.15:6379@16379 master - 0 1550566161902 2 connected 5917-10922
cee3a27cc27635da54d94f16f6375cd4acfe6c30 10.42.0.16:6379@16379 slave 505f3e126882c0c5115885e54f9b361bc7e74b97 0 1550566162506 5 connected 246c79682e6cc78b4c2c28d0e7166baf47ecb265 10.42.2.23:6379@16379 master - 0 1550566161600 8 connected 0-453 5461-5916 10923-11376 4676f8913cdcd1e256db432531c80591ae6c5fc3 10.42.2.17:6379@16379 myself,master - 0 1550566162000 1