LVS+Heartbeat实现高可用web集群方案操作记录
之前分别介绍了LVS基础知识和Heartbeat基础知识, 今天这里简单说下LVS+Heartbeat实现高可用web集群方案的操作说明.
Heartbeat 项目是 Linux-HA 工程的一个组成部分，它实现了一个高可用集群系统。心跳服务和集群通信是高可用集群的两个关键组件，在 Heartbeat 项目里，由 heartbeat 模块实现了这两个功能。
Heartbeat的高可用集群采用的通信方式是udp协议和串口通信，而且heartbeat插件技术实现了集群间的串口、多播、广播和组播通信。它实现了HA 功能中的核心功能——心跳，将Heartbeat软件同时安装在两台服务器上，用于监视系统的状态，协调主从服务器的工作，维护系统的可用性。它能侦测服务器应用级系统软件、硬件发生的故障，及时地进行错误隔绝、恢复；通过系统监控、服务监控、IP自动迁移等技术实现在整个应用中无单点故障，简单、经济地确保重要的服务持续高可用性。 Heartbeat采用虚拟IP地址映射技术实现主从服务器的切换对客户端透明的功能。但是单一的heartbeat是无法提供健壮的服务的，所以这里结合使用lvs进行负载均衡。
LVS是Linux Virtual Server的简写, 意即Linux虚拟服务器，是一个虚拟的服务器集群系统。说到lvs就得提到ipvs (ipvsadm命令)，ipvs 是 lvs集群系统的核心软件,它的主要作用是安装在 Load Balancer 上,把发往 Virtual IP 的请求转发到 Real Server 上。
ldirectord是配合lvs作为一种健康检测机制，要不负载均衡器在节点挂掉后依然没有检测的功能。
案例架构草图如下:
1) 基本环境准备 (centos6.9系统)
172.16.60.206(eth0) HA主节点(ha-master) heartbeat, ipvsadm, ldirectord
172.16.60.207(eth0) HA备节点(ha-slave) heartbeat, ipvsadm, ldirectord
172.16.60.111 VIP地址
172.16.60.204(eth0) 后端节点1(rs-204) nginx, realserver
172.16.60.205(eth0) 后端节点2(rs-205) nginx, realserver
1) 关闭防火墙和selinux (四台节点机都操作)
[root@ha-master ~]# /etc/init.d/iptables stop
[root@ha-master ~]# setenforce 0
[root@ha-master ~]# vim /etc/sysconfig/selinux
SELINUX=disabled
2) 设置主机名和绑定hosts (两台HA节点机器都操作)
主节点操作
[root@ha-master ~]# hostname ha-master
[root@ha-master ~]# vim /etc/sysconfig/network
HOSTNAME=ha-master
[root@ha-master ~]# vim /etc/hosts
172.16.60.206 ha-master
172.16.60.207 ha-slave
备节点操作
[root@ha-slave ~]# hostname ha-slave
[root@ha-slave ~]# vim /etc/sysconfig/network
HOSTNAME=ha-slave
[root@ha-slave ~]# vim /etc/hosts
172.16.60.206 ha-master
172.16.60.207 ha-slave
3) 设置ip路由转发功能 (两台HA节点机器都操作)
[root@ha-master ~]# echo 1 > /proc/sys/net/ipv4/ip_forward
[root@ha-master ~]# vim /etc/sysctl.conf
net.ipv4.ip_forward = 1
[root@ha-master ~]# sysctl -p2) 安装和配置 Heartbeat 和LVS (两台HA节点机都操作)
1) 首先安装heartbeat (HA主备两个节点都要同样操作)
分别下载epel-release-latest-6.noarch.rpm 和 ldirectord-3.9.5-3.1.x86_64.rpm
下载地址: https://pan.baidu.com/s/1IvCDEFLCBYddalV89YvonQ
提取密码: gz53
[root@ha-master ~]# ll epel-release-latest-6.noarch.rpm
-rw-rw-r-- 1 root root 14540 Nov 5 2012 epel-release-latest-6.noarch.rpm
[root@ha-master ~]# ll ldirectord-3.9.5-3.1.x86_64.rpm
-rw-rw-r-- 1 root root 90140 Dec 24 15:54 ldirectord-3.9.5-3.1.x86_64.rpm
[root@ha-master ~]# yum install -y epel-release
[root@ha-master ~]# rpm -ivh epel-release-latest-6.noarch.rpm --force
[root@ha-master ~]# yum install -y heartbeat* libnet
[root@ha-master ~]# yum install -y ldirectord-3.9.5-3.1.x86_64.rpm ＃因为依赖比较多， 所以直接采用yum方式安装
2) 配置heartbeat (HA主备两个节点都要操作)
安装完heartbeat后系统会生成一个/etc/ha.d/目录，此目录用于存放heartbeat的有关配置文件。
Heartbeat自带配置文件的注释信息较多，在此手工编写有关配置文件，heartbeat常用配置文件有四个，分别是：
ha.cf：heartbeat主配置文件
ldirectord.cf：资源管理文件
haresources：本地资源文件
authkeys：认证文件
[root@ha-master ~]# cd /usr/share/doc/heartbeat-3.0.4/
[root@ha-master heartbeat-3.0.4]# cp authkeys ha.cf haresources /etc/ha.d/
[root@ha-master heartbeat-3.0.4]# cd /usr/share/doc/ldirectord-3.9.5
[root@ha-master ldirectord-3.9.5]# cp ldirectord.cf /etc/ha.d/
[root@ha-master ldirectord-3.9.5]# cd /etc/ha.d/
[root@ha-master ha.d]# ll
total 56
-rw-r--r-- 1 root root 645 Dec 24 21:37 authkeys
-rw-r--r-- 1 root root 10502 Dec 24 21:37 ha.cf
-rwxr-xr-x 1 root root 745 Dec 3 2013 harc
-rw-r--r-- 1 root root 5905 Dec 24 21:37 haresources
-rw-r--r-- 1 root root 8301 Dec 24 21:38 ldirectord.cf
drwxr-xr-x 2 root root 4096 Dec 24 21:28 rc.d
-rw-r--r-- 1 root root 692 Dec 3 2013 README.config
drwxr-xr-x 2 root root 4096 Dec 24 21:28 resource.d
-rw-r--r-- 1 root root 2082 Mar 24 2017 shellfuncs
3) 配置heartbeat的主配置文件ha.cf （HA主备节点配置一样）
[root@ha-master ha.d]# pwd
/etc/ha.d
[root@ha-master ha.d]# cp ha.cf ha.cf.bak
[root@ha-master ha.d]# > ha.cf
[root@ha-master ha.d]# vim ha.cf
debugfile /var/log/ha-debug
logfile /var/log/ha-log #日志存放位置
#crm yes #是否开启集群资源管理功能
logfacility local0 #记录日志等级
keepalive 2 #心跳的时间间隔，默认时间单位为秒
deadtime 5 #超出该时间间隔未收到对方节点的心跳，则认为对方已经死亡。
warntime 3 #超出该时间间隔未收到对方节点的心跳，则发出警告并记录到日志中，但此时不会切换
initdead 10 #在某些系统上，系统启动或重启之后需要经过一段时间网络才能正常工作，该选项用于解决这种情况产生的时间间隔。取值至少为deadtime的两倍。
udpport 694 #设置广播通信使用的端口，694为默认使用的端口号。
bcast eth0 # Linux指定心跳使用以太网广播方式，并在eth0上进行广播。"#"后的要完全删除，要不然要出错。
ucast eth0 172.16.60.207 #采用网卡eth0的UDP多播来组织心跳，后面跟的IP地址应该为双机中对方的IP地址!!!!!
auto_failback on #在该选项设为on的情况下，一旦主节点恢复运行，则自动获取资源并取代备用节点。off主节点恢复后变为备用节点，备用为主节点!!!!!
#stonith_host * baytech 10.0.0.3 mylogin mysecretpassword
#stonith_host ken3 rps10 /dev/ttyS1 kathy 0
#stonith_host kathy rps10 /dev/ttyS1 ken3 0
#watchdog /dev/watchdog
node ha-master #主机节点名，可通过"uname -n"查看，默认为主节点!!!!!
node ha-slave #备用机节点名，默认为次节点，要注意顺序!!!!
#ping 172.16.60.207 # 选择ping节点，选择固定路由作为节点。ping节点仅用来测试网络连接。一般选择这行ping测试就行, 下面一行注释掉.
ping_group group1 172.16.60.204 172.16.60.205 #这个地址并不是双机中的两个节点地址，而是仅仅用来测试网络的连通性. 当这两个IP 都不能ping通时，对方即开始接管资源。
respawn root /usr/lib64/heartbeat/ipfail #选配项。其中rootr表示启动ipfail进程的身份。要确保/usr/lib64/heartbeat/ipfail这个路径正确(可以用find命令搜索出来), 否则heartbeat启动失败
apiauth ipfail gid=root uid=root
============
温馨提示:
HA备节点的ha.cf文件只需要将上面配置中的ucast一行内容改为"ucast eth0 172.16.60.206" 即可, 其他配置内容和上面HA主节点的ha.cf完全一样!
4) 配置heartbeat的认证文件authkeys （HA主备节点配置必须一致）
[root@ha-master ~]# cd /etc/ha.d/
[root@ha-master ha.d]# cp authkeys authkeys.bak
[root@ha-master ha.d]# >authkeys
auth 3 #auth后面指定的数字，下一行必须作为关键字再次出现! 一共有"1", "2","3" 三行, 这里选择"3"关键字, 选择"1"和"2"关键字也行, HA主备节点必须一致!
#1 crc
#2 sha1 HI!
3 md5 Hello!
必须将该文件授权为600
[root@ha-master ha.d]# chmod 600 authkeys
[root@ha-master ha.d]# ll authkeys
-rw------- 1 root root 20 Dec 25 00:16 authkeys
5) 修改heartbeat的资源文件haresources (HA主备节点配置必须完全一致)
[root@ha-slave ha.d]# cp haresources haresources.bak
[root@ha-slave ha.d]# >haresources
[root@ha-slave ha.d]# vim haresources # 在文件结尾添加下面一行内容. 由于该文件默认全是注释,可以先清空该文件, 然后添加下面这一行内容
ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
说明:
上面设置ha-maser为主节点, 集群VIP为172.16.60.111, ipvsadm ldirectord为所要监视的应用
6) 配置heartbeat的监控文件ldirectord.cf (HA主备节点配置必须完全一致)
ldirectord，用于监控在lvs集群的真实服务。ldirectord是和heartbeat相结合的一个服务，可以作为heartbeat的一个启动服务。
所以之前要安装好heartbeat，而且配置好，再将ldrectord的配置文件复制到/etc/ha.d下，因为默认没有放到里面。
[root@ha-master ha.d]# cp ldirectord.cf ldirectord.cf.bak
[root@ha-master ha.d]# >ldirectord.cf
[root@ha-master ha.d]# vim ldirectord.cf
#GlobalDirectives
checktimeout=3 #判定realserver出错时间
checkinterval=1 #指定ldirectord在两次检查之间的间隔时间
autoreload=yes #是否自动重载配置文件
logfile="/var/log/ldirectord.log" #设定Ldirectord日志输出文件路径
#logfile="local0"
#emailalert="root@30920.cn"
#emailalertfreq=3600
#emailalertstatus=all
quiescent=no #如果一个realserver节点在checktimeout设置的时间周期内没响应，将会被踢除，中断现有客户端的连接。 设置为yes, 则出问题的realserver节点不会被踢出, 只是新的连接不能到达。
# Samplefor an http virtual service
virtual=172.16.60.111:80 #指定虚拟IP，注意在virtual这行后面的行必须缩进4个空格或以一个tab字符进行标记!! 最好是一个tab字符进行标记!!!
real=172.16.60.204:80 gate #gate为DR模式，ipip表示TUNL模式，masq表示NAT模式
real=172.16.60.205:80 gate
fallback=127.0.0.1:80 gate #当所有RS机器不能访问的时候WEB重写向地址; 即表示realserver全部失败，vip指向本机80端口
service=http #指定服务类型，这里对HTTP进行负载均衡
request="lvs_testpage.html" #请求监控地址, 即监听测试页面名称，这个页面放入后端真实realserver节点的web服务的根目录下!!!
receive="Test HA Page" #指定请求和应答字符串，也就是lvs_testpage.html的内容
#virtualhost=some.domain.com.au #虚拟服务器的名称可任意指定
scheduler=wlc #指定调度算法，这里的算法一定要和lvs脚本(/etc/sysconfig/ipvsadm)的算法一样
persistent=600 #持久链接：表示600s之内同意ip将访问同一台realserver
#netmask=255.255.255.255
protocol=tcp
checktype=connect
checkport=80
温馨提示:
配置如上，通过virtual来定义vip，接下来是定义real service的节点，fallback是当所有real挂掉后，访问请求到本机的80端口上去，一般这个页面显示服务器正在维护等界面。
service表示；调度的服务，scheduler是调度算法，protocol是定义协议，checktype是检查类型为协商，checkport就是检查的端口，也就是健康检查。
上面在/etc/ha.d/ldirectord.cf文件里定义了一个80端口的代理转发, 如果还有其他端口, 比如3306,
则只需要在下面再添加一个"virtual=172.16.60.111:3306 ...."类似上面的配置即可! 配置案例在备份的ldirectord.cf.bak文件里有.
7) 安装lvs (两台HA节点机都操作)
安装lvs依赖
[root@ha-master ~]# yum install -y libnl* popt*
查看是否加载lvs模块
[root@ha-master ~]# modprobe -l |grep ipvs
kernel/net/netfilter/ipvs/ip_vs.ko
kernel/net/netfilter/ipvs/ip_vs_rr.ko
kernel/net/netfilter/ipvs/ip_vs_wrr.ko
kernel/net/netfilter/ipvs/ip_vs_lc.ko
kernel/net/netfilter/ipvs/ip_vs_wlc.ko
kernel/net/netfilter/ipvs/ip_vs_lblc.ko
kernel/net/netfilter/ipvs/ip_vs_lblcr.ko
kernel/net/netfilter/ipvs/ip_vs_dh.ko
kernel/net/netfilter/ipvs/ip_vs_sh.ko
kernel/net/netfilter/ipvs/ip_vs_sed.ko
kernel/net/netfilter/ipvs/ip_vs_nq.ko
kernel/net/netfilter/ipvs/ip_vs_ftp.ko
kernel/net/netfilter/ipvs/ip_vs_pe_sip.ko
下载并安装LVS
[root@ha-master ~]# cd /usr/local/src/
[root@ha-master src]# unlink /usr/src/linux
[root@ha-master src]# ln -s /usr/src/kernels/2.6.32-431.5.1.el6.x86_64/ /usr/src/linux
[root@ha-master src]# wget http://www.linuxvirtualserver.org/software/kernel-2.6/ipvsadm-1.26.tar.gz
[root@ha-master src]# tar -zvxf ipvsadm-1.26.tar.gz
[root@ha-master src]# cd ipvsadm-1.26
[root@ha-master ipvsadm-1.26]# make && make install
LVS安装完成，查看当前LVS集群
[root@ha-master ipvsadm-1.26]# ipvsadm -L -n
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
8) 在HA主备节点上添加lvs启动脚本
[root@ha-master ha.d]# vim /etc/sysconfig/ipvsadm
-A -t 172.16.60.111:80 -s wlc -p 600
-a -t 172.16.60.111:80 -r 172.16.60.204:80 -g
-a -t 172.16.60.111:80 -r 192.168.0.205:80 -g
说明: -p 600的意思是会话保持时间为600秒，应该和ldirectord.cf文件配置一致
9) 在后端的realserver节点上编写LVS启动脚本 (两个后端realserver操作完全一致)
[root@rs-204 ~]# vim /etc/init.d/realserver
#!/bin/sh
VIP=172.16.60.111
. /etc/rc.d/init.d/functions
case "$1" in
# 禁用本地的ARP请求、绑定本地回环地址
start)
/sbin/ifconfig lo down
/sbin/ifconfig lo up
echo "1" >/proc/sys/net/ipv4/conf/lo/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/lo/arp_announce
echo "1" >/proc/sys/net/ipv4/conf/all/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/all/arp_announce
/sbin/sysctl -p >/dev/null 2>&1
/sbin/ifconfig lo:0 $VIP netmask 255.255.255.255 up
/sbin/route add -host $VIP dev lo:0
echo "LVS-DR real server starts successfully.\n"
;;
stop)
/sbin/ifconfig lo:0 down
/sbin/route del $VIP >/dev/null 2>&1
echo "1" >/proc/sys/net/ipv4/conf/lo/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/lo/arp_announce
echo "1" >/proc/sys/net/ipv4/conf/all/arp_ignore
echo "2" >/proc/sys/net/ipv4/conf/all/arp_announce
echo "LVS-DR real server stopped.\n"
;;
status)
isLoOn=`/sbin/ifconfig lo:0 | grep "$VIP"`
isRoOn=`/bin/netstat -rn | grep "$VIP"`
if [ "$isLoON" == "" -a "$isRoOn" == "" ]; then
echo "LVS-DR real server has run yet."
else
echo "LVS-DR real server is running."
fi
exit 3
;;
*)
echo "Usage: $0 {start|stop|status}"
exit 1
esac
exit 0
启动两台realserver节点的realserver脚本
[root@rs-204 ~]# chmod 755 /etc/init.d/realserver
[root@rs-204 ~]# ll /etc/init.d/realserver
-rwxr-xr-x 1 root root 1278 Dec 24 13:40 /etc/init.d/realserver
[root@rs-204 ~]# /etc/init.d/realserver start
LVS-DR real server starts successfully.\n
设置开机启动
[root@rs-204 ~]# echo "/etc/init.d/realserver" >> /etc/rc.local
查看, 发现两台realserver节点上的lo:0上已经配置了vip地址
[root@rs-204 ~]# ifconfig
...........
lo:0 Link encap:Local Loopback
inet addr:172.16.60.111 Mask:255.255.255.255
UP LOOPBACK RUNNING MTU:65536 Metric:1
接着部署两台realserver的web测试环境 (两个后端realserver操作完全一致)
采用yum方式安装nginx (先安装nginx的yum源)
[root@rs-204 ~]# rpm -ivh http://nginx.org/packages/centos/6/noarch/RPMS/nginx-release-centos-6-0.el6.ngx.noarch.rpm
[root@rs-204 ~]# yum install nginx
realserver01的nginx配置
[root@rs-204 ~]# cd /etc/nginx/conf.d/
[root@rs-204 conf.d]# cat default.conf
[root@rs-204 conf.d]# >/usr/share/nginx/html/index.html
[root@rs-204 conf.d]# vim /usr/share/nginx/html/index.html
this is test page of realserver01:172.16.60.204
[root@rs-204 conf.d]# vim /usr/share/nginx/html/lvs_testpage.html
Test HA Page
[root@rs-204 conf.d]# /etc/init.d/nginx start
Starting nginx: [ OK ]
[root@rs-204 conf.d]# lsof -i:80
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
nginx 31944 root 6u IPv4 91208 0t0 TCP *:http (LISTEN)
nginx 31945 nginx 6u IPv4 91208 0t0 TCP *:http (LISTEN)
realserver02的nginx配置
[root@rs-205 src]# cd /etc/nginx/conf.d/
[root@rs-205 conf.d]# cat default.conf
[root@rs-205 conf.d]# >/usr/share/nginx/html/index.html
[root@rs-205 conf.d]# vim /usr/share/nginx/html/index.html
this is test page of realserver02:172.16.60.205
[root@rs-205 conf.d]# /etc/init.d/nginx start
Starting nginx: [ OK ]
[root@rs-205 conf.d]# lsof -i:80
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
nginx 20839 root 6u IPv4 289527645 0t0 TCP *:http (LISTEN)
nginx 20840 nginx 6u IPv4 289527645 0t0 TCP *:http (LISTEN)
最后分别访问realserver01和realserver02节点的nginx,:
访问http://172.16.60.204/, 出现信息"this is test page of realserver01:172.16.60.204"
访问http://172.16.60.205/, 出现信息"this is test page of realserver02:172.16.60.205"
10) 接着启动两台HA主备节点的lvs和heartbeat服务
先启动HA主节点的heartbeat,
[root@ha-master ~]# service heartbeat start
Starting High-Availability services: ERROR: Unable to find cidr_netmask.
INFO: [findif] failed
INFO: Resource is stopped
Done.
[root@ha-master ~]# ps -ef|grep heartbeat
root 6947 1 0 01:03 ? 00:00:00 heartbeat: master control process
root 6952 6947 0 01:03 ? 00:00:00 heartbeat: FIFO reader
root 6953 6947 0 01:03 ? 00:00:00 heartbeat: write: bcast eth0
root 6954 6947 0 01:03 ? 00:00:00 heartbeat: read: bcast eth0
root 6955 6947 0 01:03 ? 00:00:00 heartbeat: write: ucast eth0
root 6956 6947 0 01:03 ? 00:00:00 heartbeat: read: ucast eth0
root 6957 6947 0 01:03 ? 00:00:00 heartbeat: write: ping_group group1
root 6958 6947 0 01:03 ? 00:00:00 heartbeat: read: ping_group group1
root 6961 2247 0 01:03 pts/0 00:00:00 grep heartbeat
heartbeat服务端口默认是694. 且heartbeat服务启动时会去启动ldirectord
[root@ha-master ~]# lsof -i:694
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
heartbeat 8298 root 7u IPv4 18888 0t0 UDP *:ha-cluster
heartbeat 8299 root 7u IPv4 18888 0t0 UDP *:ha-cluster
heartbeat 8300 root 7u IPv4 18894 0t0 UDP *:ha-cluster
heartbeat 8301 root 7u IPv4 18894 0t0 UDP *:ha-cluster
查看HA主节点的heartbeat的log日志
[root@ha-master ~]# tail -f /var/log/ha-log
.........
.........
ResourceManager(default)[4001]: 2018/12/25_01:13:29 info: Acquiring resource group: ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[4029]: 2018/12/25_01:13:29 INFO: Resource is stopped
ResourceManager(default)[4001]: 2018/12/25_01:13:29 info: Running /etc/ha.d/resource.d/IPaddr 172.16.60.111 start
.........
IPaddr(IPaddr_172.16.60.111)[4125]: 2018/12/25_01:13:29 INFO: Adding inet address 172.16.60.111/24 with broadcast address 172.16.60.255 to device eth0
..........
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[4111]: 2018/12/25_01:13:29 INFO: Success
ResourceManager(default)[4001]: 2018/12/25_01:13:29 info: Running /etc/init.d/ipvsadm start
ResourceManager(default)[4001]: 2018/12/25_01:13:29 info: Running /etc/init.d/ldirectord start
查看HA主节点, 发现此时vip资源在主节点上
[root@ha-master ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:50:9b brd ff:ff:ff:ff:ff:ff
inet 172.16.60.206/24 brd 172.16.60.255 scope global eth0
inet 172.16.60.111/24 brd 172.16.60.255 scope global secondary eth0
inet6 fe80::250:56ff:feac:509b/64 scope link
valid_lft forever preferred_lft forever
在HA主备节点上可以查看到lvs负载情况
[root@ha-master ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 172.16.60.111:80 wlc persistent 600
-> 172.16.60.204:80 Route 1 0 0
-> 192.168.0.205:80 Route 1 0 0
再启动HA备节点的heartbeat
[root@ha-slave ha.d]# service heartbeat start
Starting High-Availability services: INFO: Resource is stopped
Done.
[root@ha-slave ha.d]# ps -ef|grep heartbeat
root 4257 1 0 01:04 ? 00:00:00 heartbeat: master control process
root 4260 4257 0 01:04 ? 00:00:00 heartbeat: FIFO reader
root 4261 4257 0 01:04 ? 00:00:00 heartbeat: write: bcast eth0
root 4262 4257 0 01:04 ? 00:00:00 heartbeat: read: bcast eth0
root 4263 4257 0 01:04 ? 00:00:00 heartbeat: write: ucast eth0
root 4264 4257 0 01:04 ? 00:00:00 heartbeat: read: ucast eth0
root 4265 4257 0 01:04 ? 00:00:00 heartbeat: write: ping_group group1
root 4266 4257 0 01:04 ? 00:00:00 heartbeat: read: ping_group group1
root 4269 4257 0 01:04 ? 00:00:00 /usr/lib64/heartbeat/ipfail
root 4272 4257 0 01:04 ? 00:00:00 heartbeat: master control process
查看HA备份节点, 发现此时vip资源不在备份节点上
[root@ha-slave ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:05:b5 brd ff:ff:ff:ff:ff:ff
inet 172.16.60.207/24 brd 172.16.60.255 scope global eth0
inet6 fe80::250:56ff:feac:5b5/64 scope link
valid_lft forever preferred_lft forever
在备份节点上查看不到lvs负载情况
[root@ha-slave ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn3) 故障转移切换测试
1) 首先关闭HA主节点的heartbeat服务
[root@ha-master ~]# service heartbeat stop
Stopping High-Availability services: Done.
[root@ha-master ~]# ps -ef|grep heartbeat
root 5377 3353 0 01:22 pts/0 00:00:00 grep heartbeat
查看HA主节点的heartbeat的log日志
[root@ha-master ~]# tail -f /var/log/ha-log
.........
ResourceManager(default)[5223]: 2018/12/25_01:22:46 info: Releasing resource group: ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
ResourceManager(default)[5223]: 2018/12/25_01:22:46 info: Running /etc/init.d/ldirectord stop
ResourceManager(default)[5223]: 2018/12/25_01:22:46 info: Running /etc/init.d/ipvsadm stop
ResourceManager(default)[5223]: 2018/12/25_01:22:46 info: Running /etc/ha.d/resource.d/IPaddr 172.16.60.111 stop
...........
Dec 25 01:22:49 ha-master heartbeat: [4846]: info: killing HBWRITE process 4856 with signal 15
Dec 25 01:22:49 ha-master heartbeat: [4846]: info: killing HBREAD process 4857 with signal 15
Dec 25 01:22:49 ha-master heartbeat: [4846]: info: Core process 4854 exited. 7 remaining
.........
Dec 25 01:22:49 ha-master heartbeat: [4846]: info: ha-master Heartbeat shutdown complete.
然后发现HA主节点的vip资源已经被转移走了(转移到备节点上了), 在主节点上也查看不到lvs的负载情况
[root@ha-master ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:50:9b brd ff:ff:ff:ff:ff:ff
inet 172.16.60.206/24 brd 172.16.60.255 scope global eth0
inet6 fe80::250:56ff:feac:509b/64 scope link
valid_lft forever preferred_lft forever
[root@ha-master ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
========================================================
接着到HA备节点上查看, 发现vip资源已经转移过来了, 并且在备份节点上可以查看到lvs负载情况
[root@ha-slave ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:05:b5 brd ff:ff:ff:ff:ff:ff
inet 172.16.60.207/24 brd 172.16.60.255 scope global eth0
inet 172.16.60.111/24 brd 172.16.60.255 scope global secondary eth0
inet6 fe80::250:56ff:feac:5b5/64 scope link
valid_lft forever preferred_lft forever
[root@ha-slave ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 172.16.60.111:80 wlc persistent 600
-> 172.16.60.204:80 Route 1 0 0
-> 192.168.0.205:80 Route 1 0 0
查看HA备节点的heartbeat的log日志
[root@ha-slave ~]# tail -2000 /var/log/ha-log
...........
ResourceManager(default)[7413]: 2018/12/25_01:18:15 info: Acquiring resource group: ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[7441]: 2018/12/25_01:18:15 INFO: Resource is stopped
ResourceManager(default)[7413]: 2018/12/25_01:18:15 info: Running /etc/ha.d/resource.d/IPaddr 172.16.60.111 start
IPaddr(IPaddr_172.16.60.111)[7537]: 2018/12/25_01:18:15 INFO: Adding inet address 172.16.60.111/24 with broadcast address 172.16.60.255 to device eth0
IPaddr(IPaddr_172.16.60.111)[7537]: 2018/12/25_01:18:15 INFO: Bringing device eth0 up
IPaddr(IPaddr_172.16.60.111)[7537]: 2018/12/25_01:18:15 INFO: /usr/libexec/heartbeat/send_arp -i 200 -r 5 -p /var/run/resource-agents/send_arp-172.16.60.111 eth0 172.16.60.111 auto not_used not_used
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[7523]: 2018/12/25_01:18:15 INFO: Success
ResourceManager(default)[7413]: 2018/12/25_01:18:15 info: Running /etc/init.d/ipvsadm start
ResourceManager(default)[7413]: 2018/12/25_01:18:15 info: Running /etc/init.d/ldirectord start
mach_down(default)[7386]: 2018/12/25_01:18:15 info: /usr/share/heartbeat/mach_down: nice_failback: foreign resources acquired
mach_down(default)[7386]: 2018/12/25_01:18:15 info: mach_down takeover complete for node ha-master.
2) 接着第一步测试结果, 再启动HA主节点的heartbeat服务
[root@ha-master ~]# ps -ef|grep heartbeat
root 5391 3353 0 01:25 pts/0 00:00:00 grep heartbeat
[root@ha-master ~]# service heartbeat start
Starting High-Availability services: INFO: Resource is stopped
Done.
[root@ha-master ~]# ps -ef|grep heartbeat
root 5519 1 0 01:25 ? 00:00:00 heartbeat: master control process
root 5522 5519 0 01:25 ? 00:00:00 heartbeat: FIFO reader
root 5523 5519 0 01:25 ? 00:00:00 heartbeat: write: bcast eth0
root 5524 5519 0 01:25 ? 00:00:00 heartbeat: read: bcast eth0
root 5525 5519 0 01:25 ? 00:00:00 heartbeat: write: ucast eth0
root 5526 5519 0 01:25 ? 00:00:00 heartbeat: read: ucast eth0
root 5527 5519 0 01:25 ? 00:00:00 heartbeat: write: ping_group group1
root 5528 5519 0 01:25 ? 00:00:00 heartbeat: read: ping_group group1
root 5531 3353 0 01:25 pts/0 00:00:00 grep heartbeat
查看HA主节点的heartbeat的log日志
[root@ha-master ~]# tail -f /var/log/ha-log
........
ResourceManager(default)[5566]: 2018/12/25_01:25:58 info: Acquiring resource group: ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[5594]: 2018/12/25_01:25:58 INFO: Resource is stopped
ResourceManager(default)[5566]: 2018/12/25_01:25:58 info: Running /etc/ha.d/resource.d/IPaddr 172.16.60.111 start
IPaddr(IPaddr_172.16.60.111)[5690]: 2018/12/25_01:25:58 INFO: Adding inet address 172.16.60.111/24 with broadcast address 172.16.60.255 to device eth0
IPaddr(IPaddr_172.16.60.111)[5690]: 2018/12/25_01:25:58 INFO: Bringing device eth0 up
IPaddr(IPaddr_172.16.60.111)[5690]: 2018/12/25_01:25:58 INFO: /usr/libexec/heartbeat/send_arp -i 200 -r 5 -p /var/run/resource-agents/send_arp-172.16.60.111 eth0 172.16.60.111 auto not_used not_used
/usr/lib/ocf/resource.d//heartbeat/IPaddr(IPaddr_172.16.60.111)[5676]: 2018/12/25_01:25:58 INFO: Success
ResourceManager(default)[5566]: 2018/12/25_01:25:58 info: Running /etc/init.d/ipvsadm start
ResourceManager(default)[5566]: 2018/12/25_01:25:58 info: Running /etc/init.d/ldirectord star
......
然后发现HA主节点的heartbeat服务恢复后, vip资源又抢回来了, 并且又能在主节点上查看lvs负载情况
因为在ha.cf 文件里配置了"auto_failback on", 表示 一旦主节点恢复运行，则自动获取资源并取代备用节点!
[root@ha-master ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:50:9b brd ff:ff:ff:ff:ff:ff
inet 172.16.60.206/24 brd 172.16.60.255 scope global eth0
inet 172.16.60.111/24 brd 172.16.60.255 scope global secondary eth0
inet6 fe80::250:56ff:feac:509b/64 scope link
valid_lft forever preferred_lft forever
[root@ha-master ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 172.16.60.111:80 wlc persistent 600
-> 172.16.60.204:80 Route 1 0 0
-> 192.168.0.205:80 Route 1 0 0
============================================================
接着到HA备节点上查看, 发现vip资源已经被转移走了(又转移回主节点上了), 并且在备份节点上查看不到lvs负载情况
[root@ha-slave ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
link/ether 00:50:56:ac:05:b5 brd ff:ff:ff:ff:ff:ff
inet 172.16.60.207/24 brd 172.16.60.255 scope global eth0
inet6 fe80::250:56ff:feac:5b5/64 scope link
valid_lft forever preferred_lft forever
[root@ha-slave ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
查看HA备节点的heartbeat的log日志
[root@ha-slave ~]# tail -2000 /var/log/ha-log
.........
ResourceManager(default)[8376]: 2018/12/25_01:25:58 info: Releasing resource group: ha-master IPaddr::172.16.60.111 ipvsadm ldirectord
ResourceManager(default)[8376]: 2018/12/25_01:25:58 info: Running /etc/init.d/ldirectord stop
ResourceManager(default)[8376]: 2018/12/25_01:25:58 info: Running /etc/init.d/ipvsadm stop
ResourceManager(default)[8376]: 2018/12/25_01:25:58 info: Running /etc/ha.d/resource.d/IPaddr 172.16.60.111 stop
3) 访问http://172.16.60.111
.........
关闭realserver两个节点中的任何一个nginx, 比如关闭rs-204的nginx
则访问lvs, 即http://172.16.60.111不受影响
查看lvs负载情况
[root@ha-master ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 172.16.60.111:80 wlc persistent 600
-> 192.168.0.205:80 Route 1 0 0
说明rs-204节点已经从lvs集群中踢出来了
接着恢复rs-204的nginx, 再次查看lvs情况
[root@ha-master ~]# ipvsadm -Ln
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP 172.16.60.111:80 wlc persistent 600
-> 172.16.60.204:80 Route 1 0 0
-> 192.168.0.205:80 Route 1 0 0
说明rs-204节点的nginx恢复后, 再次被加到lvs集群中.
这是因为在ldirectord.cf文件中设置了quiescent=no
整个故障过程, 对于http://172.16.60.111 的lvs代理层的访问没有任何影响!本文参与 腾讯云自媒体同步曝光计划，分享自作者个人站点/博客。
原始发表：2018-12-24 ，如有侵权请联系 cloudcommunity@tencent.com 删除
腾讯云开发者
Copyright © 2013 - 2026 Tencent Cloud. All Rights Reserved. 腾讯云 版权所有
深圳市腾讯计算机系统有限公司 ICP备案/许可证号：粤B2-20090059 
粤公网安备44030502008569号
腾讯云计算（北京）有限责任公司 京ICP证150476号 | 京ICP备11018762号