LVS http 访问VIP时不能完成三次握手

hellogzd

问题是这样的，我一共用了三台服务器，其中一台是Direct Server，其余两台是Real server。按照网上介绍的方法，进行配置。结果最后访问http://虚拟IP 总是出现网页无法访问的情况，不能转到实际的服务器上。而通过ipvsadm命令却可以看到http访问已经平均分发到两台服务器上了，即inActConn数随着访问页面的增多，也增多。

请各位大侠给支个招。

hellogzd

Directory Server的piranha 配置如下：
serial_no = 22
primary = 192.168.0.4
service = lvs
backup = 0.0.0.0
heartbeat = 1
heartbeat_port = 539
keepalive = 6
deadtime = 18
network = direct
debug_level = NONE
virtual www.nxist.com {
     active = 1
     address = 192.168.0.1 eth0:0
     vip_nmask = 255.255.255.255
     port = 80
     send = "GET / HTTP/1.0\r\n\r\n"
     expect = "HTTP"
     use_regex = 0
     load_monitor = none
     scheduler = rr
     protocol = tcp
     timeout = 6
     reentry = 15
     quiesce_server = 0
     server node1.nxist.com {
         address = 192.168.0.3
         active = 1
         port = 80
         weight = 1
     }
     server node2.nxist.com {
         address = 192.168.0.5
         active = 1
         port = 80
         weight = 1
     }
     }
Real Server执行脚本如下：
#!/bin/bash
VIP=192.168.0.1
/etc/rc.d/init.d/functions
/sbin/ifconfig lo:0 $VIP broadcast $VIP netmask 255.255.255.255 up
/sbin/route add -host $VIP 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
#end

南非蚂蚁

192.168.0.3和192.168.0.5能直接访问服务吗，另外贴出日志

hellogzd

192.168.0.3和192.168.0.5都可以直接访问，倘若关闭Director Server上的pulse服务，在Director Server上也可访问VIP，日志没有报错信息，但可以看到192.168.0.3:80 available之类的信息。wireshark可以在Real Server上抓到客户端访问http的tcp包。
重启pulse过程中日志如下：
Apr  2 21:25:59 osnagios pulse[16349]: Terminating due to signal 15
Apr  2 21:25:59 osnagios lvs[16416]: shutting down due to signal 15
Apr  2 21:25:59 osnagios lvs[16416]: shutting down virtual service www.nxist.com
Apr  2 21:25:59 osnagios nanny[16425]: Terminating due to signal 15
Apr  2 21:25:59 osnagios nanny[16426]: Terminating due to signal 15
Apr  2 21:25:59 osnagios avahi-daemon[3847]: Withdrawing address record for 192.168.0.1 on eth0.
Apr  2 21:26:00 osnagios pulse[16549]: STARTING PULSE AS MASTER
Apr  2 21:26:10 osnagios pulse[16549]: partner dead: activating lvs
Apr  2 21:26:10 osnagios lvs[16606]: starting virtual service www.nxist.com active: 80
Apr  2 21:26:10 osnagios nanny[16610]: starting LVS client monitor for 192.168.0.1:80
Apr  2 21:26:10 osnagios nanny[16610]: [ active ] making 192.168.0.3:80 available
Apr  2 21:26:10 osnagios lvs[16606]: create_monitor for www.nxist.com/osnode1.nxist.com running as pid 16610
Apr  2 21:26:10 osnagios nanny[16612]: starting LVS client monitor for 192.168.0.1:80
Apr  2 21:26:10 osnagios lvs[16606]: create_monitor for www.nxist.com/osnode2.nxist.com running as pid 16612
Apr  2 21:26:10 osnagios avahi-daemon[3847]: Registering new address record for 192.168.0.1 on eth0.
Apr  2 21:26:10 osnagios avahi-daemon[3847]: Withdrawing address record for 192.168.0.1 on eth0.
Apr  2 21:26:10 osnagios avahi-daemon[3847]: Registering new address record for 192.168.0.1 on eth0.
Apr  2 21:26:10 osnagios avahi-daemon[3847]: Withdrawing address record for 192.168.0.1 on eth0.
Apr  2 21:26:10 osnagios avahi-daemon[3847]: Registering new address record for 192.168.0.1 on eth0.
Apr  2 21:26:10 osnagios nanny[16612]: [ active ] making 192.168.0.5:80 available
Apr  2 21:26:15 osnagios pulse[16614]: gratuitous lvs arps finished
在10.1.1.2上telnet 192.168.0.1 80，监控到如下信息：
Every 0.1s: ipvsadm -L                                  Mon Apr  2 21:36:58 2012

IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddressort Scheduler Flags
  -> RemoteAddressort           Forward Weight ActiveConn InActConn
TCP  192.168.0.1:http rr
  -> 192.168.0.5:http                 Route     1          0                 1
  -> 192.168.0.3:http                 Route     1          0                 1
注：lvs负载均衡器的eth1:10.1.1.1，访问客户IP10.1.1.2,10.1.1.2可以ping同192.168.0.3及192.168.0.5

longbillliu

我遇到的问题和你的一模一样，不知道你解决没有？我的还没有解决

hellogzd

我的拓扑图搞错了，我采用的是DR模式，Dirctor Server、Real Server 及客户端应该放在同一个网络中。