LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

[lvs-users] Why tcpdump capture 2 duplicated UDP packet when use LVS-DR

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: [lvs-users] Why tcpdump capture 2 duplicated UDP packet when use LVS-DR for UDP Server
From: linbo liao <llbgurs@xxxxxxxxx>
Date: Fri, 23 Dec 2016 18:04:54 +0800
Hi,

I want to setup a LVS cluster to loadbalance UDP server. The test setup use
an echo UDP server, the setup:


*# Environment*

1. LVS Director:  172.24.143.181 Redhat 6.6 x86_64
2. RS: 172.24.143.182  Redhat 6.6 x86_64
3. VIP 172.24.143.187
4. ipvs 1.2.1
5. keepalived v1.2.23
6. test client: 172.24.143.183  Redhat 6.6 x86_64

*# Configure*

keepalived.conf

virtual_ipaddress {
>         172.24.143.187 dev eth1
>     }

.....
>
virtual_server 172.24.143.187 3333 {
>     delay_loop 6
>     lb_algo rr
>     lb_kind DR
>     protocol UDP
>
>     real_server 172.24.143.182 3333 {
>         weight 1
>     }
> }
>

$ ipvsadm -Ln

> UDP  172.24.143.187:3333 rr
>   -> 172.24.143.182:3333          Route   1      0          0


*# UDP server*

RS run UDP server
http://svn.python.org/projects/python/trunk/Demo/sockets/udpecho.py

*s.bind(('172.24.143.187', port))  # udp listen VIP address*
>

$ python udpserver.py -s 3333



*# Test*$ nc -u 172.24.143.187 3333
job
job


*# Tcpdump on LVS director  (2 packet are duplicated, but one is
padding)*18:03:33.778558
In 00:50:56:83:68:bd ethertype IPv4 (0x0800), length 62: (tos 0x0, ttl 64,
id 46680, offset 0, flags [DF], proto UDP (17), length 32)
    172.24.143.183.56006 > 172.24.143.187.3333: [udp sum ok] UDP, length 4
    0x0000:  0000 0001 0006 0050 5683 68bd 0000 0800  .......PV.h.....
    0x0010:  4500 0020 b658 4000 4011 0cd1 ac18 8fb7  E....X@.@.......
    0x0020:  ac18 8fbb dac6 0d05 000c d3ec 6a6f 620a  ............job.
    0x0030:  0000 0000 0000 0000 0000 0000 0000       ..............
18:03:33.778598 Out 00:50:56:83:06:ef ethertype IPv4 (0x0800), length 48:
(tos 0x0, ttl 64, id 46680, offset 0, flags [DF], proto UDP (17), length 32)
    172.24.143.183.56006 > 172.24.143.187.3333: [udp sum ok] UDP, length 4
    0x0000:  0004 0001 0006 0050 5683 06ef 0000 0800  .......PV.......
    0x0010:  4500 0020 b658 4000 4011 0cd1 ac18 8fb7  E....X@.@.......
    0x0020:  ac18 8fbb dac6 0d05 000c d3ec 6a6f 620a  ............job.


*# Tcpdump from client (the output does make sence, one is outgoing packet,
one is incoming packet)*18:03:33.776888 Out 00:50:56:83:68:bd ethertype
IPv4 (0x0800), length 48: (tos 0x0, ttl 64, id 46680, offset 0, flags [DF],
proto UDP (17), length 32)
    172.24.143.183.56006 > 172.24.143.187.3333: [bad udp cksum 2b5c!] UDP,
length 4
        0x0000:  0004 0001 0006 0050 5683 68bd 0000 0800  .......PV.h.....
        0x0010:  4500 0020 b658 4000 4011 0cd1 ac18 8fb7  E....X@.@.......
        0x0020:  ac18 8fbb dac6 0d05 000c 77c1 6a6f 620a  ..........w.job.
18:03:33.778472  In 00:50:56:83:6b:cd ethertype IPv4 (0x0800), length 62:
(tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto UDP (17), length 32)
    172.24.143.187.3333 > 172.24.143.183.56006: [udp sum ok] UDP, length 4
        0x0000:  0000 0001 0006 0050 5683 6bcd 0000 0800  .......PV.k.....
        0x0010:  4500 0020 0000 4000 4011 c329 ac18 8fbb  E.....@.@..)....
        0x0020:  ac18 8fb7 0d05 dac6 000c d3ec 6a6f 620a  ............job.
        0x0030:  0000 0000 0000 0000 0000 0000 0000       ..............

Why two same packet show on LVS director, any one know the reason?

Thanks,
Linbo
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/

LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users

<Prev in Thread] Current Thread [Next in Thread>