RiCo技術農場

我之前是在windows建立redis cluster with sentinel，

我得老實說，config針對多環境且多instance還滿痛苦的，

我常常都會被搞混又浪費大半天再找錯誤，透過人眼掃描並設定確實耗時費力，

這時透過docker統一處理絕對是首選，而且也可以輕鬆實現IaC，

我將透過Dockerfile建立base image with redis，

並透過docker-compose來建立一套redis cluster with sentinel服務，

同時，我也可以輕鬆又簡單達到scale out目的，我把之前架構修改如下

Dockerfile

FROM redis:4
MAINTAINER RiCo
RUN chown redis:redis /data
VOLUME /data
WORKDIR /data
RUN mkdir -p /usr/local/etc/redis && chmod ugo=+rwx /usr/local/etc/redis
COPY redis-master.conf /usr/local/etc/redis/redis.conf
EXPOSE 6379
CMD ["redis-server","/usr/local/etc/redis/redis.conf"]

 

Dockerfile-sentinel

FROM redis:4
MAINTAINER RiCo
RUN chown redis:redis /data
VOLUME /data
WORKDIR /data
COPY sentinel.conf /usr/local/etc/redis/sentinel.conf
RUN chmod -R 777 /usr/local/etc/redis 
ENV SENTINEL_QUORUM 2
ENV SENTINEL_DOWN_AFTER 5000
ENV SENTINEL_FAILOVER 8000
ENV SENTINEL_PORT 26379
EXPOSE 26379
COPY entrypoint.sh /usr/local/bin/entrypoint.sh
RUN chmod -R 777 /usr/local/bin/entrypoint.sh
ENTRYPOINT ["/usr/local/bin/entrypoint.sh"]

 

docker-compose.yml

# docker-compose up -d --build
version: '3'

services:

  master:
      image: redis:4
      build:
         context: .
         dockerfile: Dockerfile
      volumes:
       - ./redis/master:/data
      restart: unless-stopped
  slave:
    image: redis:4
    build:
         context: .
         dockerfile: Dockerfile
    command: redis-server --slaveof redis-master 6379
    volumes:
     - ./redis/slave:/data
    restart: unless-stopped
    depends_on:
      - master
    links:
      - master:redis-master
  sentinel:
    image: redis:4-sentinel
    build: 
         context: . #./sentinel/
         dockerfile: Dockerfile-sentinel
    volumes:
     - ./redis/sentinel:/tmp
    depends_on:
      - master
    environment:
     - SENTINEL_DOWN_AFTER=5000
     - SENTINEL_FAILOVER=5000
    links:
     - master:redis-master
     - slave

 

redis.conf 和sentinel.conf 可以依照個人需求修改，我針對integration環境設定幾個參數如下

maxmemory 500mb
bind 0.0.0.0
maxmemory-policy allkeys-lru
protected-mode no
appendonly yes
logfile "redis_log"
syslog-enabled yes

 

@執行docker-compose up –d

@scale out slave

Sudo docker-compose scale slave=3

用不到3秒，我們立馬有3個slave’s instance。

 

@scale out sentinel

Sudo docker-compose scale sentinel=3

現在我們有3個sentinel’s instance，而且相關設定都參考之前base來取代對應環境變數，一整個輕鬆愜意。

 

@確認

當我們進入master redis，透過redis-cli > info，可以看到slave數量=3

我先暫停master service，來看看自動failover是否可以正常working

Sudo docker-compose pause master

等達到所設定“DOWN-AFTER” and “FAIL AFTER”條件後，

我們可以查看sentinel log確認是否有自動failover到另一台新的master

Sudo docker-compose logs sentinel

可以看到新的master已經變更為 172.18.0.6，最後我們再恢復原本master。

Sudo docker-compose unpause master

 

 

參考

Redis(6)-設定Redis Sentinel & Replication

5 ways to debug an exploding Docker container

/bin/sh^M : bad interpreter

Redis Clustering With Redis Sentinel On Docker

redis-cluster-with-sentinel