2

On Debian 64bit, I configured a lxc like this :

host ip : 172.16.0.25 vm guest on this host : 172.16.0.44

I've done the bridge on my host

auto eth0 
iface eth0 inet manual
        up ifconfig eth0 0.0.0.0 up
        down ifconfig eth0 down

auto br0 
iface br0 inet static
        bridge_ports eth0
        bridge_fd       0
        address 172.16.0.25
        netmask 255.255.255.0
        gateway 172.16.0.52

On guest side, I changed the config with

lxc.utsname = myGuest
lxc.network.type = veth
lxc.network.flags = up
lxc.network.link = br0
lxc.network.name = eth0
lxc.network.hwaddr = 00:FF:12:34:56:78
lxc.network.ipv4 = 172.16.0.44/24

When I run the VM : my host (172.16.0.25) and my guest (172.16.0.44) can ping each other But now, other physical machines on the same network (172.16.0.0) cannot see my guest.

Same issue from my guest that can only see the IP of its host !

In fact I would like to get the same behaviour than a bridge network on VmWare Workstation. What kind of routage VmWare does to be seen "everywhere" on the network ?

(By the way, my host is already on a ESX server, but I think it is not relevant at this level)

Improve this question
1
  • You should set tcpdump and see what's going on when requests are being made. At this point, I would suspect arp requests not being relayed from host to guests. At least, a layer 2 problem, as there should not be routing issue (same subnet). Is your guest able to ping hosts outside its own host? Is there any firewalling issue?
    – philippe
    Commented Jan 29, 2013 at 13:06

1 Answer 1

Reset to default
4

Your host is virtualized with VMware, then it is plugged to the virtual switch which does not expect any more mac-addresses to be shown than those of VMware virtualized servers (so not lxc inside the VMware guest). So this is definitely a layer 2 issue: VMWare (at least VSwitch) received paquets with a mac-address different than the host mac-address and drop them. So there would be no way for the guest to reach outside its host, as it should go through the Vswitch to do that.

A workaround is to enabled different mac-addresses on the Vswitch. It is performed by editing settings on the Vswitch (on the VMware host, configuration, networking, select the vswitch your lxc-host is connected to, properties, Edit, security, MAC address changes -> Accept).

Improve this answer
10
  • Great post Philippe, But The "Accept" flag is already on...
    – Stef
    Commented Jan 29, 2013 at 13:39
  • mmmh thanks! but I don't see any accept flag anywhere, or do I have missed it? ;)
    – philippe
    Commented Jan 29, 2013 at 13:42
  • Ok sorry, the "MAC address changes" of the vswitch0 is already checked with "Accept" value
    – Stef
    Commented Jan 29, 2013 at 13:53
  • Sorry! I may not have told you the appropriate flag to set on; It may be Promiscuous mode (the Vswitch then acts like a dummy network hub - be careful to security issue then, VMware guests would be able to get the traffic of the whole subnet); I am not quite sure about which of these options have to be set up :/ Hope it will help you finding a solution!
    – philippe
    Commented Jan 29, 2013 at 14:02
  • Ok, cool advice. At the end, What you said is if I do all the "stuff" on "physical machine" I will not have the problem don't you ?
    – Stef
    Commented Jan 29, 2013 at 14:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .