linuxea:lvs-dr4层代理 我们试图将后端的业务通过lvs转到nginx代理，nginx分别作为4成和7层代理，这里使用DR模型，那就意味着只能进行做4层代理。类似如：tomcat，mq等也可以进行代理，试图减少代码的耦合度，将他们拆分，使用一个vpi加端口的形式。我准备了4台机器加一台redis做测试lvs 10.10.240.144 and 10.10.240.143 nginx-proxy 10.10.240.113 And 10.10.240.114 redis 10.10.240.145 VIP 10.10.240.188安装lvs+keepalived[root@linuxea-vm-Node_10_10_240_144 ~]# yum install keepalived ipvsadm -y修改keepavlied配置文件/etc/keepalived/keepalived.conf ，添加vip 10.10.240.188,已经后端的代理节点的端口以及Ip配置信息[root@linuxea-vm-Node_10_10_240_144 /etc/keepalived]$ cat /etc/keepalived/keepalived.conf ! Configuration File for keepalived global_defs { router_id LVS_DEVEL } vrrp_instance platformtransfer { state MASTER interface eth0 virtual_router_id 63 priority 99 advert_int 1 authentication { auth_type PASS auth_pass 1111 } virtual_ipaddress { 10.10.240.188/16 brd 10.10.255.255 dev eth0 label eth0:trans1 } } virtual_server 10.10.240.188 880 { delay_loop 6 lb_algo rr lb_kind DR protocol TCP nopreempt garp_master_delay 10 real_server 10.10.240.114 880 { weight 1 TCP_CHECK { connect_timeout 3 nb_get_retry 3 delay_before_retry 3 connect_port 880 } } real_server 10.10.240.113 880 { weight 1 TCP_CHECK { connect_timeout 3 nb_get_retry 3 delay_before_retry 3 connect_port 880 } } } virtual_server 10.10.240.188 6379 { delay_loop 6 lb_algo rr lb_kind DR protocol TCP nopreempt garp_master_delay 10 real_server 10.10.240.114 6379 { weight 1 TCP_CHECK { connect_timeout 3 nb_get_retry 3 delay_before_retry 3 connect_port 6379 } } real_server 10.10.240.113 6379 { weight 1 TCP_CHECK { connect_timeout 3 nb_get_retry 3 delay_before_retry 3 connect_port 6379 } } }添加lvs脚本/scripts/lvs.sh [root@linuxea-vm-Node_10_10_240_144 /etc/keepalived]$ cat /scripts/lvs.sh #!/bin/bash #注：下文中LVS_IP_INF变量中的信息为需要配置的需求变量，变量之间分别以@符合隔开，总共有7个变量。以本例的需求来解析： #lvs节点通过vip 10.10.240.188,将880，6379端口以DR模型的方式转发到10.10.240.114和113节点的880，6379端口 #[-g|i|m]: LVS类型 # -g: DR # -i: TUN # -m: NAT #本脚本中默认的调度模式为rr #通过以上解析，第一个变量绑定vip的网卡，第二个变量为vip，第三个变量为lvs-RS转发节点（可多个，以空格隔开）， #第四个变量为lvs-RS节点端口号，第五个变量为该业务备注信息，第六个变量为lvs节点中使用的端口号，第七个变量为lvs的类型 LVS_IP_INF=`cat << EOF eth0@10.10.240.188@10.10.240.114 10.10.240.113@880@test880@880@g eth0@10.10.240.188@10.10.240.114 10.10.240.113@6379@redis6379@6379@g EOF` case "$1" in start) /usr/sbin/ipvsadm -C echo "$LVS_IP_INF" | while read line;do read NET_FACE VIP PROJ SPORT MODE < <(echo "$line" | awk -F"@" '{print $1 " " $2 " " $5 " " $6 " " $7}') RIPS=$(echo $line | awk -F"@" '{print $3}') PORTS=$(echo $line | awk -F"@" '{print $4}') echo "添加项目LVS ${PROJ}: 网卡--- ${NET_FACE} 虚拟IP--- ${VIP} 真实主机--- ${RIPS} 代理端口--- ${PORTS}" for port in ${PORTS};do echo "添加虚拟服务器记录 ipvsadm -At ${VIP}:${SPORT} -s rr" /usr/sbin/ipvsadm -At ${VIP}:${SPORT} -s rr for rip in ${RIPS};do echo "添加真实服务器记录 ipvsadm -at ${VIP}:${SPORT} -r ${rip}:${port} -${MODE}" /usr/sbin/ipvsadm -at ${VIP}:${SPORT} -r ${rip}:${port} -${MODE} done echo done echo done echo "当前LVS状态：" /usr/sbin/ipvsadm -Ln ;; stop) /usr/sbin/ipvsadm -C /usr/sbin/ipvsadm -Ln ;; add) date echo "$LVS_IP_INF" | while read line;do read NET_FACE VIP PROJ SPORT MODE < <(echo "$line" | awk -F"@" '{print $1 " " $2 " " $5 " " $6 " " $7}') RIPS=`echo $(echo $line | awk -F"@" '{print $3}')` PORTS=`echo $(echo $line | awk -F"@" '{print $4}')` echo "添加LVS项目 ${PROJ}: 网卡---${NET_FACE} 虚拟IP---${VIP} 真实主机---${RIPS} 代理端口---${PORTS}" for port in ${PORTS};do /usr/sbin/ipvsadm -Ln | grep -v '-' | grep ${VIP}:${SPORT} > /dev/null [ $? -eq 0 ] && echo $VIP:${SPORT}"已存在" && continue echo "添加虚拟服务器记录 ipvsadm -At ${VIP}:${SPORT} -s rr" /usr/sbin/ipvsadm -At ${VIP}:${SPORT} -s rr for rip in ${RIPS};do echo "添加真实服务器记录 ipvsadm -at ${VIP}:${SPORT} -r ${rip}:${port} -${MODE}" /usr/sbin/ipvsadm -at ${VIP}:${SPORT} -r ${rip}:${port} -${MODE} done echo done echo done echo "当前LVS状态：" /usr/sbin/ipvsadm -Ln ;; *) echo "Usage: $0 {start|stop|add}" ;; esac而后需要添加x执行权限chmod +x /scripts/lvs.sh 配置systemd风格的启动脚本/usr/lib/systemd/system/addvip.service and /usr/lib/systemd/system/keepalived.service cat /usr/lib/systemd/system/addvip.service [Unit] Description=Add lvs vip After=keepalived.service multi-user.target [Service] Type=oneshot RemainAfterExit=yes ExecStart=/scripts/lvs.sh start ExecStop=/scripts/lvs.sh stop [Install] WantedBy=multi-user.target cat /usr/lib/systemd/system/keepalived.service [Unit] Description=LVS and VRRP High Availability Monitor After=syslog.target network-online.target [Service] Type=forking PIDFile=/var/run/keepalived.pid KillMode=process EnvironmentFile=-/etc/sysconfig/keepalived ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS ExecReload=/bin/kill -HUP $MAINPID [Install] WantedBy=multi-user.target放行112端口，两台keepaliaved之间-A INPUT -p 112 -s 10.10.240.143 -j ACCEPT -A INPUT -p tcp -m tcp -m state --state NEW -m multiport --dports 880,6379 -m comment --comment "lvs" -j ACCEPT-A INPUT -p 112 -s 10.10.240.144 -j ACCEPT -A INPUT -p tcp -m tcp -m state --state NEW -m multiport --dports 880,6379 -m comment --comment "lvs" -j ACCEPT启动systemctl start keepalived.service addvip.service启动后的ip会进行绑定到keepalived的master节点之上[root@linuxea-vm-Node_10_10_240_144 /etc/keepalived]$ ip a 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000 link/ether 88:88:2f:f0:48:19 brd ff:ff:ff:ff:ff:ff inet 10.10.240.144/8 brd 10.255.255.255 scope global dynamic eth0 valid_lft 84948sec preferred_lft 84948sec inet 10.10.240.188/16 brd 172.25.255.255 scope global eth0:trans1 valid_lft forever preferred_lft forever[root@linuxea-vm-Node_10_10_240_144 /etc/keepalived]$ ipvsadm -L -n IP Virtual Server version 1.2.1 (size=4096) Prot LocalAddress:Port Scheduler Flags -> RemoteAddress:Port Forward Weight ActiveConn InActConn TCP 10.10.240.188:880 rr -> 10.10.240.113:880 Route 1 0 0 -> 10.10.240.114:880 Route 1 0 0 TCP 10.10.240.188:6379 rr -> 10.10.240.113:6379 Route 1 0 0 -> 10.10.240.114:6379 Route 1 0 0 配置4层负载配置4层nginx代理后端的redis节点，便于测试，仅仅配置了一台redis，在配置nginx之前，需要将vip绑定到nginx层的lo网络接口上，如下：[root@linuxea-vm-Node113 /etc/nginx/stream]# cat /scripts/lvs.sh #!/bin/bash ######################################################################### # File Name: /scripts/lvs.sh # Author: LookBack # Email: admin#dwhd.org # Version: # Created Time: 2018年10月13日 星期六 10时20分41秒 ######################################################################### VIP1=10.10.240.188 case "$1" in start) ip addr add ${VIP1}/32 brd $VIP1 dev lo label lo:0 ip route add $VIP1 dev lo:0 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 sysctl -p >/dev/null 2>&1 echo "RealServer Start OK" ;; stop) ip addr del ${VIP1}/32 brd $VIP1 dev lo label lo:0 ip route del $VIP1 dev lo:0 echo "0" >/proc/sys/net/ipv4/conf/lo/arp_ignore echo "0" >/proc/sys/net/ipv4/conf/lo/arp_announce echo "0" >/proc/sys/net/ipv4/conf/all/arp_ignore echo "0" >/proc/sys/net/ipv4/conf/all/arp_announce echo "RealServer Stoped" ;; *) echo "Usage: $0 {start|stop}" ;; esac两台都需要绑定，而后如下：[root@linuxea-vm-Node113 /etc/nginx/stream]# ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet 10.10.188.188/32 brd 10.10.188.188 scope global lo:0 valid_lft forever preferred_lft forever inet 10.10.240.188/32 brd 10.10.240.188 scope global lo:0 valid_lft forever preferred_lft forever[root@linuxea-vm-Node114 ~]# ip a 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet 10.10.188.188/32 brd 10.10.188.188 scope global lo:0 valid_lft forever preferred_lft forever inet 10.10.240.188/32 brd 10.10.240.188 scope global lo:0 valid_lft forever preferred_lft forever并且配置放行从lvs到nginx的端口-A INPUT -s 10.10.240.0/24 -p tcp -m tcp -m state --state NEW -m multiport --dports 880,6379 -j ACCEPT配置nginx[root@linuxea-vm-Node113 ~]# cat docker-compose-nginx_vts.yml version: '3' services: nginx_vts: image: marksugar/nginx:v1.14.0-vts container_name: nginx restart: always network_mode: "host" volumes: - /etc/nginx:/etc/nginx/ - /data/:/data/ environment: - NGINXCONF=on - NGINX_PORT=80 - SERVER_NAME=www.linuxea.net - PHP_FPM_SERVER=127.0.0.1:9000 ports: - "80"修改添加/etc/nginx/nginx.conf一个字段stream { include stream/*.conf; }而后添加redis节点的upstream[root@linuxea-vm-Node113 ~]# cat /etc/nginx/stream/redis.conf upstream 6379 { server 10.10.240.145:6379; } server { listen 6379; proxy_pass 6379; }配置redisredis使用之前的编写的docker配置[root@linuxea-vm-Node_10_10_240_145 /data1/redis]$ cat docker-compose-redis-4-0-11.yml version: '2' services: redis: image: marksugar/redis:4.0.11 container_name: redis restart: always network_mode: "host" privileged: true environment: - REDIS_CONF=on - REQUIREPASSWD=OTdmOWI4ZTM4NTY1M2M4OTZh - MASTERAUTHPAD=OTdmOWI4ZTM4NTY1M2M4OTZh - MAXCLIENTS_NUM=600 - MAXMEMORY_SIZE=4096 volumes: - /etc/localtime:/etc/localtime:ro - /etc/redis:/etc/redis - /data/redis-data:/data/redis:Z - /data/logs:/data/logs测试在redis本机访问lvs dr的vip代理的redis ip和端口。[root@linuxea-vm-Node_10_10_240_145 /data1/redis]$ redis-cli -h 10.10.240.188 -p 6379 -a OTdmOWI4ZTM4NTY1M2M4OTZh info |grep cpu used_cpu_sys:805.85 used_cpu_user:747.25 used_cpu_sys_children:0.00 used_cpu_user_children:0.00

lvs负载均衡之lvs_dr(notes二) lvs-dr：直接路由 Director在实现转发时不修改请求的IP首部，而是通过直接封装MAC首部完成转发；目标MAC是Director根据调度方法挑选出某RS的MAC地址；拓扑结构有别有NAT类型； 架构特性： (1) 保证前端路由器将目标地址为VIP的请求报文通过ARP地址解析后送往Director 解决方案： 静态绑定：在前端路由直接将VIP对应的目标MAC静态配置为Director的MAC地址； arptables：在各RS上，通过arptables规则拒绝其响应对VIP的ARP广播请求； 内核参数：在RS上修改内核参数，并结合地址的配置方式实现拒绝响应对VIP的ARP广播请求； (2) RS的RIP可以使用私有地址；但也可以使用公网地址，此时可通过互联网上的主机直接对此RS发起管理操作； (3) 请求报文必须经由Director调度，但响应报文必须不能经由Director; (4) 各RIP必须与DIP在同一个物理网络中； (5) 不支持端口映射； (6) RS可以使用大多数的OS； (7) RS的网关一定不能指向Director； (1) 各RS要直接响应Client，因此，各RS均得配置VIP；但仅能够让Director上的VIP能够与本地路由直接通信； (2) Director不会拆除或修改请求报文的IP首部，而是通过封闭新的帧首部（源MAC为Director的MAC，目标MAC为挑选出的RS的MAC）完成调度； 2.4.26, 2.6.4 kernel引入了两个内核参数： arp_announce：定义arp通知级别； arp_ignore：定义arp忽略arp请求或arp通告的级别； /proc/sys/net/ipv4/conf/INTERFACE node3 在/proc/sys/net/ipv4/conf下 1，修改两个接口，all和lo [root@node3 conf]# echo 1 > /proc/sys/net/ipv4/conf/all/arp_ignore [root@node3 conf]# echo 1 > /proc/sys/net/ipv4/conf/lo/arp_ignore [root@node3 conf]# echo 2 > /proc/sys/net/ipv4/conf/all/arp_announce [root@node3 conf]# echo 2 > /proc/sys/net/ipv4/conf/lo/arp_announce [root@node3 conf]# sysctl -a | grep arp net.ipv4.conf.lo.arp_announce = 2 net.ipv4.conf.lo.arp_ignore = 1 2，添加lo:0地址， [root@node3 ~]# ifconfig lo:0 172.16.100.202 netmask 255.255.255.255 broadcast 172.16.100.202 up [root@node3 ~]# ifconfig lo:0 lo:0 Link encap:Local Loopback inet addr:172.16.100.202 Mask:255.255.255.255 UP LOOPBACK RUNNING MTU:65536 Metric:1 3，添加路由，如果你访问的地址是172.16.100.202,要求必须通过lo:0进出 [root@node3 ~]route add -host 172.16.100.202 dev lo:0 4，启动httpd,创建测试页面 node2 1，修改两个接口，all和lo [root@node2 ~]# echo 1 > /proc/sys/net/ipv4/conf/all/arp_ignore [root@node2 ~]# echo 1 > /proc/sys/net/ipv4/conf/lo/arp_ignore [root@node2 ~]# echo 2 > /proc/sys/net/ipv4/conf/all/arp_announce [root@node2 ~]# echo 2 > /proc/sys/net/ipv4/conf/lo/arp_announce [root@node2 ~]# sysctl -a | grep arp net.ipv4.conf.lo.arp_announce = 2 net.ipv4.conf.lo.arp_ignore = 1 2，添加lo:0地址 [root@node2 ~]# ifconfig lo:0 172.16.100.202 netmask 255.255.255.255 broadcast 172.16.100.202 up [root@node2 ~]# ifconfig lo:0 lo:0 Link encap:Local Loopback inet addr:172.16.100.202 Mask:255.255.255.255 UP LOOPBACK RUNNING MTU:65536 Metric:1 3，添加路由，如果你访问的地址是172.16.100.202,要求必须通过lo:0进出 [root@node2 ~]route add -host 172.16.100.202 dev lo:0 4，启动httpd,创建测试页面 node1 1，添加eht1:0地址，（仅对自己访问，广播域broadcast限制） root@node1 ~]# ifconfig eth1:0 172.16.100.202 netmask 255.255.255.255 broadcast 172.16.100.202 up [root@node1 ~]# ifconfig eth1:0 eth1:0 Link encap:Ethernet HWaddr 00:0C:29:78:10:11 inet addr:172.16.100.202 Bcast:172.16.100.202 Mask:255.255.255.255 2，清空ipvsadm配置 [root@node1 ~]# ipvsadm -C 3，清空iptables配置 [root@node1 ~]# iptables -F 4，给本机添加路由指向 [root@node1 ~]# route add -host 172.16.100.202 dev eth1:0 5，开启网卡转发 [root@node1 ~]# sysctl -a |grep ip_forward net.ipv4.ip_forward = 1 6，添加集群服务，-s指明调度方法rr [root@node1 ~]# ipvsadm -A -t 172.16.100.202:80 -s rr 7，添加172.16.100.202:80的客户端服务器172.16.249.157,类型为-g DR,权重为5 [root@node1 ~]# ipvsadm -a -t 172.16.100.202:80 -r 172.16.249.157 -g -w 5 8，添加172.16.100.202:80的客户端服务器172.16.249.186,类型为-g DR,权重为1 [root@node1 ~]# ipvsadm -a -t 172.16.100.202:80 -r 172.16.249.186 -g -w 1 其实在dr中，权重是无任何意义的 [root@node1 ~]# ipvsadm -Ln IP Virtual Server version 1.2.1 (size=4096) Prot LocalAddress:Port Scheduler Flags -> RemoteAddress:Port Forward Weight ActiveConn InActConn TCP 172.16.100.202:80 rr -> 172.16.249.157:80 Route 5 0 0 -> 172.16.249.186:80 Route 1 0 0 [root@node1 ~]# 客户端： 在客户机arp -a 查看两个ip地址的mac地址 172.16.100.202 00-0c-29-78-10-11 172.16.249.117 00-0c-29-78-10-11 浏览器测试     不同ip直接构建，需要现将网络打通，其他便简单 [root@node3 ~]# ifconfig eth2 192.168.0.101/24 up [root@node3 ~]# route add default gw 192.168.0.254 [root@node2 ~]# ifconfig eth2 192.168.0.102/24 up [root@node2 ~]# route add default gw 192.168.0.254 [root@node1 ~]# ifconfig eth1 192.168.0.100/24 up [root@node1 ~]# ipvsadm -A -t 172.16.100.202:80 -s rr [root@node1 ~]# ipvsadm -a -t 172.16.100.202:80 -r 192.168.0.101 -g -w 1 [root@node1 ~]# ipvsadm -a -t 172.16.100.202:80 -r 192.168.0.102 -g -w 3 [root@node1 ~]# ipvsadm -Ln IP Virtual Server version 1.2.1 (size=4096) Prot LocalAddress:Port Scheduler Flags -> RemoteAddress:Port Forward Weight ActiveConn InActConn TCP 172.16.100.202:80 rr -> 192.168.0.101:80 Route 1 0 0 -> 192.168.0.102:80 Route 3 0 0 [root@node1 ~]#