OpenVZ Forum: Support » Package loss in virtual machine (don not have arp-reply)

Home » General » Support » Package loss in virtual machine (don not have arp-reply)

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Package loss in virtual machine (don not have arp-reply) [message #43600]

Fri, 30 September 2011 07:59

epiphany
Messages: 2
Registered: September 2011

Junior Member

Hi.
I have problem in virtual machine with arp-reply.

Configuration my system:
host: SL6.1 x86_64
vzkernel-2.6.32-042stab037.1.x86_64
vzctl-3.0.29.3-1.x86_64
vzquota-3.0.12-1.x86_64
vzkernel-2.6.32-042stab036.1.x86_64
vzkernel-firmware-2.6.32-042stab037.1.noarch
vzctl-lib-3.0.29.3-1.x86_64

vz.conf default and edit parametr to NEIGHBOUR_DEVS=all

virtual: SL6.1 x86_64 template

host have 2 interface:

eth0:192.168.1.10
eth1:88.88.88.88

virtual machine have 2 interface:

eth0:192.168.1.11
eth1:88.88.88.89

route in host:

88.88.88.89 dev veth1.0  scope link 
192.168.1.11 dev veth1.1  scope link 
88.88.88.88/29 dev eth1  proto kernel  scope link  src 88.88.88.88 
192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.10 
192.168.0.0/16 via 192.168.1.1 dev eth0 
default via 88.88.88.87 dev eth1

route in virtual machine:

192.168.0.0/16 dev eth1  scope link 
default dev eth0  scope link

From host machine to ping in external (like 8.8.8.8 or 8.8.4.4) and internal (like gateway 192.168.1.1) resoures without packet loss all time.
From virtaul machine to ping external resourse don't have problem, but in internal resourse - have packet loss like this

--- 192.168.1.1 ping statistics ---
100 packets transmitted, 34 received, 66% packet loss, time 100001ms
rtt min/avg/max/mdev = 0.462/10.894/352.395/59.447 ms, pipe 3

In this time from host machive to this local address 192.168.1.1 don't have packet loss.
If i shutdown external interface in virtual machine - packet loss to 192.168.1.1 are terminated, if up - begins again.

In moment when ping not loss tcpdump like this

11:54:43.228466 ARP, Request who-has 192.168.1.1 tell 192.168.1.11, length 28
11:54:43.228496 ARP, Reply 192.168.1.1 is-at 00:18:51:6c:52:2e, length 28
11:54:43.228529 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 46, length 64
11:54:43.229127 IP 192.168.1.1 > 192.168.1.11: ICMP echo reply, id 29186, seq 46, length 64
11:54:44.228493 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 47, length 64
11:54:44.229142 IP 192.168.1.1 > 192.168.1.11: ICMP echo reply, id 29186, seq 47, length 64
11:54:45.228498 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 48, length 64
11:54:45.228995 IP 192.168.1.1 > 192.168.1.11: ICMP echo reply, id 29186, seq 48, length 64

And then ping is loss - like this

11:54:23.229405 ARP, Request who-has 192.168.1.11 tell 88.88.88.88, length 28
11:54:24.228465 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 27, length 64
11:54:24.229402 ARP, Request who-has 192.168.1.11 tell 88.88.88.88, length 28
11:54:25.228467 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 28, length 64
11:54:25.229407 ARP, Request who-has 192.168.1.11 tell 88.88.88.88, length 28
11:54:26.228469 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 29, length 64
11:54:27.228464 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 30, length 64
11:54:27.229414 ARP, Request who-has 192.168.1.11 tell 88.88.88.88, length 28
11:54:28.228464 IP 192.168.1.11 > 192.168.1.1: ICMP echo request, id 29186, seq 31, length 64

If i drop arp-table in virtual machine (ip neig flush all), ping already begins to go.

This problem are observed in kernel 2.6.32-042stab037.1, 2.6.32-042stab036.6.

Report message to a moderator