环境准备：
| IP | 版本 | 角色 |
|---|---|---|
| 172.x.x.11 | RHEL 7.9 | master |
| 172.x.x.12 | RHEL 7.9 | replica |
| 172.x.x.13 | RHEL 7.9 | replica |
安装docker-ce：
# 移除旧版本
sudo yum remove docker \
docker-client \
docker-client-latest \
docker-common \
docker-latest \
docker-latest-logrotate \
docker-logrotate \
docker-engine
# 配置仓库
sudo yum install -y yum-utils device-mapper-persistent-data lvm2
#官方源
#sudo yum-config-manager --add-repo https://download.docker.com/linux/centos/docker-ce.repo
#阿里云源
sudo yum-config-manager --add-repo https://mirrors.aliyun.com/docker-ce/linux/centos/docker-ce.repo
sudo yum makecache fast
# 检查可安装的版本
yum list docker-ce --showduplicates | sort -r
# 安装指定版本
#sudo yum install docker-ce- docker-ce-cli- containerd.io docker-buildx-plugin docker-compose-plugin
yum install -y docker-ce-25.0.1 docker-ce-cli-25.0.1 containerd.io docker-buildx-plugin docker-compose-plugin
# 启动
sudo systemctl enable docker
sudo systemctl start docker
配置阿里云镜像源:
cat > /etc/docker/daemon.json << EOF
{
"registry-mirrors" : ["https://b9pmyelo.mirror.aliyuncs.com"]
}
EOF
sudo systemctl restart docker
🐬Docker社区版部署参见：
- https://docs.docker.com/engine/install/centos/
- https://gottdeskrieges.blog.csdn.net/article/details/113242029
安装docker-compose：
curl -SL https://github.com/docker/compose/releases/download/v2.24.6/docker-compose-linux-x86_64 -o /usr/local/bin/docker-compose
chmod 755 /usr/local/bin/docker-compose
📖参考：https://docs.docker.com/compose/install/standalone/
- Redis版本：7.2.4
- Redis安装路径：/opt/docker-compose/redis
准备单台服务的compose文件，每台服务器上部署一个redis实例、一个sentinel实例和一个redis exporter。容器网络采用仅主机模式，避免NAT转发的复杂性。
version: '3'
services:
redis-server:
image: redis:7.2.4
network_mode: host
container_name: redis-server
restart: unless-stopped
#ports:
# - 6379:6379
environment:
TZ: "Asia/Shanghai"
command: ["/conf/redis.conf"]
volumes:
- ./conf/:/conf/
- ./data/:/data/
redis-sentinel:
image: redis:7.2.4
network_mode: host
container_name: redis-sentinel
restart: unless-stopped
environment:
TZ: "Asia/Shanghai"
#ports:
# - 26379:26379
command: ["/conf/sentinel.conf","--sentinel"]
volumes:
- ./conf/:/conf/
redis-exporter:
image: oliver006/redis_exporter
network_mode: host
restart: unless-stopped
environment:
TZ: Asia/Shanghai
REDIS_ADDR: redis://localhost:6379
REDIS_USER: redis_monitor
REDIS_PASSWORD: Monpass_xxxx
# ports:
# - 9121:9121
如果没有现成的redis exporter镜像，可以采用本地部署，并注释掉docker compose文件中的对应内容。
直接下载二进制安装包：
cd /opt
wget https://github.com/oliver006/redis_exporter/releases/download/v1.58.0/redis_exporter-v1.58.0.linux-amd64.tar.gz
tar -xvf redis_exporter-v1.58.0.linux-amd64.tar.gz
cd redis_exporter
./redis_exporter --redis.addr=redis://localhost:6379 --redis.user=redis_monitor --redis.password="Monpass_xxxx"
检查redis exporter采集到的数据：
curl -X GET http://localhost:9121/metrics
📖参考：https://github.com/oliver006/redis_exporter
哨兵节点配置文件/opt/docker-compose/redis/conf/sentinel.conf：
#port 26379
loglevel verbose
sentinel announce-ip ""
sentinel announce-port 26379
dir "/data"
sentinel monitor 6379 2
sentinel auth-user sentinel_user
sentinel auth-pass xxxxxx
sentinel down-after-milliseconds 15000
其中，
数据库节点配置文件/opt/docker-compose/redis/conf/redis.conf：
bind 0.0.0.0
#port 6379
timeout 3600
tcp-keepalive 300
loglevel notice
databases 16
save 600 100
save 120 10000
save 7200 1
dir /data
masteruser # 主从复制用户
masterauth
replica-read-only yes
#min-replicas-to-write 3
#min-replicas-max-lag 10
replica-announce-ip ""
replica-announce-port 6379
aclfile /conf/users.acl
maxclients 10000
maxmemory 10gb
maxmemory-policy noeviction
appendonly yes
appendfsync everysec
auto-aof-rewrite-percentage 100
auto-aof-rewrite-min-size 64mb
配置好管理员用户、监控用户、sentinel用户、主从复制用户、以及应用用户。
在redis.conf中配置aclfile参数，ACL文件（conf/users.acl）示例如下：
user default off nopass &* -@all
user root on >Rootpass_xxxx allkeys allchannels +@all
user redis_monitor on >Monpass_xxxx +client +ping +info +config|get +cluster|info +slowlog +latency +memory +select +get +scan +xinfo +type +pfcount +strlen +llen +scard +zcard +hlen +xlen +eval allkeys
user sentinel_user on >Senpass_xxxx allchannels +multi +slaveof +ping +exec +subscribe +config|rewrite +role +publish +info +client|setname +client|kill +script|kill
user replica_user on >Replpass_xxxx +psync +replconf +ping
user appuser on >Apppass_xxxx allkeys allchannels +@all -@admin -@dangerous
导入ACL文件配置：
# 列出已有的用户权限清单
docker exec -it redis-server redis-cli acl list
# 导入ACL文件中的用户权限配置
docker exec -it redis-server redis-cli acl load
docker exec -it redis-server redis-cli acl list
📖参考：https://redis.io/docs/management/security/acl/#create-and-edit-user-acls-with-the-acl-setuser-command
修改Linux内核参数：
sed -i '/vm.overcommit_memory/d' /etc/sysctl.conf
echo "vm.overcommit_memory = 1" >> /etc/sysctl.conf
echo "net.core.somaxconn=2048" >> /etc/sysctl.conf
sysctl -p
禁用透明大页：
cat <<EOF >> /etc/rc.d/rc.local
if test -f /sys/kernel/mm/transparent_hugepage/enabled; then
echo never > /sys/kernel/mm/transparent_hugepage/enabled
fi
EOF
chmod +x /etc/rc.d/rc.local
编辑/etc/security/limits.conf和/etc/security/limits.d/90-nproc.conf，添加以下配置：
* soft core unlimited
* hard core unlimited
* soft nofile 65536
* hard nofile 65536
* soft nproc 65536
* hard nproc 65536
最后重启服务器生效。
修改文件权限：
cd /opt/docker-compose/
chown -R polkitd:root redis/
在docker-compose.yml文件所在路径下执行以下命令来启停单个节点上的REDIS实例。
#启动容器
cd /opt/docker-compose/redis && docker-compose up -d
#停止容器
cd /opt/docker-compose/redis && docker-compose stop
将当前节点redis实例配置为172.x.x.11上的实例的从实例：
docker exec -it redis-server redis-cli --user xx -a xxxxxx replicaof 172.x.x.11 6379
也可以把replicaof 172.x.x.11 6379配置到从实例的redis.conf文件中。
检查各节点角色和同步状态：
docker exec -it redis-server redis-cli --user xx -a xxxxxx -h 172.x.x.11 -p 6379 info replication
docker exec -it redis-server redis-cli --user xx -a xxxxxx -h 172.x.x.12 -p 6379 info replication
docker exec -it redis-server redis-cli --user xx -a xxxxxx -h 172.x.x.13 -p 6379 info replication
测试主从切换：
# 停止172.x.x.11上的主实例
docker stop redis-server
# 检查哨兵节点日志输出
docker logs -fn100 redis-sentinel
# 输出以下日志内容表示发生了故障转移
...
1:X 01 Jun 2020 17:31:50.435 # +sdown master mymaster 172.x.x.11 6379 # ->主实例客观宕机
1:X 01 Jun 2020 17:31:50.458 # +new-epoch 4
1:X 01 Jun 2020 17:31:50.460 # +vote-for-leader 79c1f8276a8fedd78a9f987da8c85fce0ea7f751 4
1:X 01 Jun 2020 17:31:50.527 # +odown master mymaster 172.x.x.11 6379 #quorum 3/2 # ->主实例主观宕机
1:X 01 Jun 2020 17:31:50.527 # Next failover delay: I will not start a failover before Mon Jun 1 17:37:50 2020
1:X 01 Jun 2020 17:31:51.708 # +config-update-from sentinel 79c1f8276a8fedd78a9f987da8c85fce0ea7f751 172.x.x.12 26379 @ mymaster 172.x.x.11 6379
1:X 01 Jun 2020 17:31:51.708 # +switch-master mymaster 172.x.x.11 6379 172.x.x.12 6379 # ->主实例地址发生切换
1:X 01 Jun 2020 17:31:51.708 * +slave slave 172.x.x.13:6379 172.x.x.13 6379 @ mymaster 172.x.x.12 6379
1:X 01 Jun 2020 17:31:51.708 * +slave slave 172.x.x.11:6379 172.x.x.11 6379 @ mymaster 172.x.x.12 6379
...
上面的信息表示主实例从172.x.x.11切换到了172.x.x.12。
主从切换后，主从节点的sentinel.conf中的replicaof参数会发生变化，相应的epoch参数也会自动+1。
References：
【1】https://www.codenong.com/cs106478941/
【2】https://lipeng1667.github.io/2021/06/21/set-up-redis-with-replication-and-sentinel-using-docker-compose/