Zabbix监控Redis服务
OpenFalcon监控Redis服务
CacheCloud搭建
系统CentOS 7.4，mysql 5.7，Redis 4.0
# yum -y install git
# cd /usr/local/
# git clone https://github.com/sohutv/cachecloud.git
卸载OpenJdk
# rpm -qa | grep jdk
# rpm -e --nodeps ...
# yum -y remove ...
安装Java环境
# tar xvf jdk-8u65-linux-x64.tar.gz -C /usr/local
配置环境变量
export JAVA_HOME=/usr/local/jdk1.8.0_65
export JAVA_BIN=/usr/local/jdk1.8.0_65/bin
export PATH=$PATH:$JAVA_HOME/bin
export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar
# source /etc/profile
# java -version
安装MySQL5.7略
初始化CacheCloud数据
innodb_page_size = 16384
sql_mode = ""
mysql> create database cachecloud charset utf8;
mysql> use cachecloud;
mysql> source /usr/local/cachecloud/script/cachecloud.sql;
创建CacheCloud连接用户
mysql> grant all on cachecloud.* to 'admin'@'%' identified by 'admin';
mysql> flush privileges;
CacheCloud使用了maven作为项目构建的工具，所以先安装maven
# yum -y install maven（注意安装maven时会安装openjdk）
CacheCloud提供了local.properties和online.properties两套配置作为测试、线上的隔离
# vim /usr/local/cachecloud/cachecloud-open-web/src/main/swap/local.properties
cachecloud.db.url = jdbc:mysql://192.168.1.101:3306/cachecloud
cachecloud.db.user = admin
cachecloud.db.password = admin
cachecloud.maxPoolSize = 20
isClustered = true
isDebug = true
spring-file = classpath:spring/spring-local.xml
log_base = /opt/cachecloud-web/logs
web.port = 9999
log.level = INFO
# vim /usr/local/cachecloud/cachecloud-open-web/src/main/swap/online.properties
cachecloud.db.url = jdbc:mysql://192.168.1.101:3306/cachecloud
cachecloud.db.user = admin
cachecloud.db.password = admin
cachecloud.maxPoolSize = 20
isClustered = true
isDebug = false
spring-file=classpath:spring/spring-online.xml
log_base=/opt/cachecloud-web/logs
web.port=8585
log.level=WARN
创建快照
在cachecloud根目录下运行
# mvn clean compile install -Plocal
在cachecloud-open-web模块下运行
# mvn spring-boot:run
生产环境部署
在cachecloud根目录下运行
# mvn clean compile install -Ponline
执行deploy.sh脚本：
# cd /usr/local/cachecloud/script/
# bash deploy.sh /usr/local/
安装maven时会安装openjdk，会导致跟你安装的oracle jdk有冲突，所以编译完CacheCloud后可以卸载openjdk
# yum -y remove java-1.8.0-openjdk*
启动CacheCloud
# bash /opt/cachecloud-web/start.sh
客户端执行
# sh /usr/local/cachecloud/script/cachecloud-init.sh
访问管理平台http://192.168.1.101:8585/
进入管理后台
机器管理
应用申请，流程审批
Redis-Sentinel,Redis-Cluster
redis-cli --stat
# gem install redis-stat
# redis-stat 192.168.1.101:6379 192.168.1.101:6380 192.168.1.102:6379 192.168.1.102:6380 192.168.1.103:6379 192.168.1.103:6380 --server=63790 10 --daemon
指定web界面的端口为63790，10秒监控一次，deamon为后台运行，访问192.168.1.101：63790
slow log,info,benchmark,monitor
监控指标
1、redis服务进程
ps aux | grep -E "redis-server.*$port"
2、连接客户数
grep "connected_clients:" ${tmpFile} | awk -F ":" '{print $2}'
3、阻塞连接数
grep "blocked_clients:" ${tmpFile} | awk -F ":" '{print $2}'
4、 redis占用内存，单位Byte转成MB
grep "used_memory:" ${tmpFile} | awk -F ":" '{print $2}' | awk '{printf "%.2f",$1/1024/1024}'
5、内存峰值，单位Byte转成MB
grep "used_memory_peak:" ${tmpFile} | awk -F ":" '{print $2}' | awk '{printf "%.2f",$1/1024/1024}'
6、主从角色
grep "role:" ${tmpFile} | awk -F ":" '{print $2}' # master（主），slave（从）
7、 master_link_status
grep "master_link_status:" ${tmpFile} | awk -F ":" '{print $2}' # up down
down: Master 已经不可访问了，Slave依然运行良好，并且保留有AOF与RDB文件
8、执行命令总数和qps
grep "total_commands_processed:" ${tmpFile} | awk -F ":" '{print $2}'
计算qps需要计算两次total_commands_processed，然后除以时间差。逻辑是第一分钟将total_commands_processed的值和当时采集该值的时间保存到last.cache中，第二分钟采集的时候获取值和时间，和上次相减得到两个差值相除即可。
9、键个数 (keys): redis实例包含的键个数