specific binds vlan to NIC

docent · Aug 20, 2009

Hi!

How can I bind a specific network card of a virtual machine to a specific network card of a host?
I'd like to bind eth0,eth1,eth2 of a VM (KVM) to eth2,eth0,eth1 of a host.

By hand, I did so:

Code:

/usr/bin/kvm \
    -monitor unix:/var/run/qemu-server/105.mon,server,nowait \
    -vnc unix:/var/run/qemu-server/105.vnc,password \
    -pidfile /var/run/qemu-server/105.pid \
    -daemonize \
    -usbdevice tablet \
    -name gw.ugk \
    -smp 2 \
    -id 105 \
    -cpuunits 1000 \
    -vga cirrus \
    -tdf \
    -k en-us \
    -drive file=/var/lib/vz/images/105/vm-105-disk.qcow2,if=ide,index=0,boot=on \
    -m 256 \
    -net tap,vlan=0,ifname=vmtab105i0,script=/var/lib/qemu-server/bridge-vlan1 \
    -net nic,vlan=0,model=rtl8139,macaddr=6E:67:B8:69:A6:74 \
    -net tap,vlan=1,ifname=vmtab105i1,script=/var/lib/qemu-server/bridge-vlan2 \
    -net nic,vlan=1,model=rtl8139,macaddr=C6:38:D5:C9:0B:3C \
    -net tap,vlan=2,ifname=vmtab105i2,script=/var/lib/qemu-server/bridge-vlan0 \
    -net nic,vlan=2,model=rtl8139,macaddr=96:D8:17:47:EF:C3

dietmar · Aug 21, 2009

You can create several bridges, and add a virtual network card on each bridge. All that can be done on the web interface.

docent · Aug 24, 2009

Not quite. In the manager I can't establish order of NICs in VM.
For example, there is only one NIC on my host - eth0. I want to create an additional two virtual NICs.

HTML:

Ethernet devices
Model   Bridge  MAC address
rtl8139 vmbr0   C6:38:D5:C9:0B:3C 
e1000   vmbr1   96:D8:17:47:EF:C3 
e1000   vmbr2   6E:67:B8:69:A6:74

In that case rtl8139 becomes eth0 in my VM, but I need that it would becomes eth2 and two e1000 become eth0 and eth1, respectively.

dietmar · Aug 24, 2009

The order is controlled by the VM, not by the host. So this depends on the VM Guest you use (for debian modify /etc/udev/rules.d/70-persistent-net.rules)

docent · Aug 24, 2009

Yes, of course, but...
I wrote this code to 70-persistent-net.rules on VM (Debian 5.02):

Code:

# PCI device (e1000)
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="96:D8:17:47:EF:C3", ATTR{type}=="1", KERNEL=="eth*", NAME="eth0"
 # PCI device (e1000)
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="6E:67:B8:69:A6:74", ATTR{type}=="1", KERNEL=="eth*", NAME="eth1"
 # PCI device 0x10ec:0x8168 (r8169)
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="C6:38:D5:C9:0B:3C", ATTR{type}=="1", KERNEL=="eth*", NAME="eth2"

However,
dmesg out:

Code:

[    4.996402] eth0: RTL-8139C+ at 0xffffc200000b8000, c6:38:d5:c9:0b:3c, IRQ 10
[    5.376181] e1000: 0000:00:04.0: e1000_probe: (PCI:33MHz:32-bit) 96:d8:17:47:ef:c3
[    5.447551] e1000: eth1: e1000_probe: Intel(R) PRO/1000 Network Connection
[    5.856766] e1000: 0000:00:05.0: e1000_probe: (PCI:33MHz:32-bit) 6e:67:b8:69:a6:74
[    5.910219] e1000: eth2: e1000_probe: Intel(R) PRO/1000 Network Connection
[    5.942697] 8139too Fast Ethernet driver 0.9.28
[   18.463286] eth0: link up, 100Mbps, full-duplex, lpa 0x05E1
[   18.841670] ADDRCONF(NETDEV_UP): eth1: link is not ready
[   18.907697] e1000: eth1: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[   18.907697] ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
[   19.192304] e1000: eth2: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[   19.199258] ADDRCONF(NETDEV_UP): eth2: link is not ready
[   19.209594] ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready

mii-tool out:

Code:

eth0: no autonegotiation, 100baseTx-FD flow-control, link ok
eth1: no autonegotiation, 1000baseT-FD flow-control, link ok
eth2: no autonegotiation, 1000baseT-FD flow-control, link ok

But, if I run VM using this command:

Code:

/usr/bin/kvm \
    -monitor unix:/var/run/qemu-server/105.mon,server,nowait \
    -vnc unix:/var/run/qemu-server/105.vnc,password \
    -pidfile /var/run/qemu-server/105.pid \
    -daemonize \
    -usbdevice tablet \
    -name gw.ugk \
    -smp 2 \
    -id 105 \
    -cpuunits 1000 \
    -vga cirrus \
    -tdf \
    -k en-us \
    -drive file=/var/lib/vz/images/105/vm-105-disk.qcow2,if=ide,index=0,boot=on \
    -m 256 \
    -net tap,vlan=0,ifname=vmtab105i0,script=/var/lib/qemu-server/bridge-vlan1 \
    -net nic,vlan=0,model=e1000,macaddr=6E:67:B8:69:A6:74 \
    -net tap,vlan=1,ifname=vmtab105i1,script=/var/lib/qemu-server/bridge-vlan2 \
    -net nic,vlan=1,model=e1000,macaddr=C6:38:D5:C9:0B:3C \
    -net tap,vlan=2,ifname=vmtab105i2,script=/var/lib/qemu-server/bridge-vlan0 \
    -net nic,vlan=2,model=rtl8139,macaddr=96:D8:17:47:EF:C3

I get the following:
dmesg out:

Code:

[    5.691851] e1000: 0000:00:03.0: e1000_probe: (PCI:33MHz:32-bit) 6e:67:b8:69:a6:74
[    5.750277] e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
[    6.156770] e1000: 0000:00:04.0: e1000_probe: (PCI:33MHz:32-bit) c6:38:d5:c9:0b:3c
[    6.218555] e1000: eth1: e1000_probe: Intel(R) PRO/1000 Network Connection
[    6.258424] eth2: RTL-8139C+ at 0xffffc200000b8000, 96:d8:17:47:ef:c3, IRQ 10
[    6.272274] 8139too Fast Ethernet driver 0.9.28
[   16.197838] e1000: eth0: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[   16.197838] ADDRCONF(NETDEV_UP): eth0: link is not ready
[   16.200839] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[   16.612507] e1000: eth1: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
[   16.986511] eth2: link up, 100Mbps, full-duplex, lpa 0x05E1

mii-tool out:

Code:

eth0: no autonegotiation, 1000baseT-FD flow-control, link ok
eth1: no autonegotiation, 1000baseT-FD flow-control, link ok
eth2: no autonegotiation, 100baseTx-FD flow-control, link ok

ie just what I need.

It would be nice if you add in the configuration file something like this:

Code:

vlan0: e1000=96:D8:17:47:EF:C3,bridge-vlan1
vlan1: e1000=6E:67:B8:69:A6:74,bridge-vlan2
vlan2: rtl8139=C6:38:D5:C9:0B:3C,bridge-vlan0

or just by order in config file.

dietmar · Aug 24, 2009

docent said:
or just by order in config file.

I thought we already do that - please check with

# qm showcmd <VMID>

docent · Aug 24, 2009

dietmar said:
I thought we already do that - please check with

# qm showcmd <VMID>

Unfortunately, no.

Code:

# qm showcmd 105
/usr/bin/kvm \
-monitor unix:/var/run/qemu-server/105.mon,server,nowait \
-vnc unix:/var/run/qemu-server/105.vnc,password \
-pidfile /var/run/qemu-server/105.pid \
-daemonize \
-usbdevice tablet \
-name gw.ugk \
-smp 2 \
-id 105 \
-cpuunits 1000 \
-vga cirrus \
-tdf \
-k en-us \
-drive file=/var/lib/vz/images/105/vm-105-disk.qcow2,if=ide,index=0,boot=on \
-m 256 \
-net tap,vlan=[B][COLOR=Red]0[/COLOR][/B],ifname=vmtab105i[COLOR=Red][B]0[/B][/COLOR],script=/var/lib/qemu-server/bridge-vlan[B][COLOR=Red]0[/COLOR][/B] \
-net nic,vlan=[COLOR=Red][B]0[/B][/COLOR],model=rtl8139,macaddr=C6:38:D5:C9:0B:3C \
-net tap,vlan=[B][COLOR=Blue]1[/COLOR][/B],ifname=vmtab105i[B][COLOR=Blue]1[/COLOR][/B],script=/var/lib/qemu-server/bridge-vlan[COLOR=Blue][B]1[/B][/COLOR] \
-net nic,vlan=[COLOR=Blue][B]1[/B][/COLOR],model=e1000,macaddr=96:D8:17:47:EF:C3 \
-net tap,vlan=[B][COLOR=Lime]2[/COLOR][/B],ifname=vmtab105i[B][COLOR=Lime]2[/COLOR][/B],script=/var/lib/qemu-server/bridge-vlan[B][COLOR=Lime]2[/COLOR][/B] \
-net nic,vlan=[COLOR=Lime][B]2[/B][/COLOR],model=e1000,macaddr=6E:67:B8:69:A6:74

You generate a command line with identical number of vlan, ifname and bridge-vlan.
QemuServer.pm:

Code:

for (my $i = 0; $i < 10; $i++) {
  if ($conf->{"vlan$i"} && (my $vlan = parse_vlan ($conf->{"vlan$i"}))) {
    $foundnet = 1;
    my $ifname = "vmtab${vmid}i[B][COLOR=Red]$i[/COLOR][/B]";
    push @$cmd, '-net', "tap,vlan=[B][COLOR=Red]$i[/COLOR][/B],ifname=$ifname,script=/var/lib/qemu-server/bridge-vlan[B][COLOR=Red]$i[/COLOR][/B]";
    foreach my $nic (@{$vlan->{nics}}) {
      push @$cmd, '-net', "nic,vlan=$i,model=$nic->{model},macaddr=$nic->{macaddr}";
    }
  }
}

dietmar · Aug 24, 2009

docent said:
You generate a command line with identical number of vlan, ifname and bridge-vlan.

Sorry, I still don't get it. What is the suggestion - reserve the order? why?

docent · Aug 25, 2009

dietmar said:
Sorry, I still don't get it. What is the suggestion - reserve the order? why?

I have only one NIC (vmbr0) on the host and two fake ifaces (vmbr1 and vmbr2) that I plan to use as internal LAN. And I need to place this NIC (vmbr0) as eth2 on VM.
I don't know why 70-persistent-net.rules does not work on VM, on hw computers it works fine.

dietmar · Aug 25, 2009

docent said:
I don't know why 70-persistent-net.rules does not work on VM, on hw computers it works fine.

It works witout problems here.

Search

Search

specific binds vlan to NIC

docent

Renowned Member

dietmar

Proxmox Staff Member

docent

Renowned Member

dietmar

Proxmox Staff Member

docent

Renowned Member

dietmar

Proxmox Staff Member

docent

Renowned Member

dietmar

Proxmox Staff Member

docent

Renowned Member

dietmar

Proxmox Staff Member