802.1q VLAN HOWTO

Doing 802.1q trunking to an Ethernet switch is really easy. I describe two methods - a text only one for servers, and a GUI one for desktops.

The VLAN number that an interface uses always gets appended to the physical interface after a dot, so for instance an interface eth0 using VLAN number 99 would be called eth0.99. The instructions below all assume you want to connect to vlan 99 - change the number as required.

Server console method
Install the vlan package:

Code:

---

sudo apt-get install vlan

---

Load the 8021q kernel module:

Code:

---

sudo modprobe 8021q

---

Ensure that this module gets loaded after a reboot by adding to /etc/modules:

Code:

---

sudo  sh -c 'grep -q 8021q /etc/modules || echo 8021q >> /etc/modules'

---

Configure the interface by editing /etc/network/interfaces:

Code:

---

sudo nano /etc/network/interfaces

---

and adding a stanza like this:

Code:

---

auto eth0.99
iface eth0.99 inet static
address 192.168.99.1
netmask 255.255.255.0

---

Then bring the interface up:

Code:

---

sudo ifup eth0.99

---

Desktop GUI method
Install the vlan package using Synaptic: System->Administration->Synaptic Package Manager
and search for and install the package 8021q.

The next bit has to be done from the command line.
Load the 8021q kernel module:

Code:

---

sudo modprobe 8021q

---

Ensure that this module gets loaded after a reboot by adding to /etc/modules:

Code:

---

sudo  sh -c 'grep -q 8021q /etc/modules || echo 8021q >> /etc/modules'

---

Now we have to create a low-level interface that the network manager can see:

Code:

---

vconfig add eth0 99

---

Now we can find eth0.99 in the GUI System->Administration0>Network and configure it how we want.

Using the above method, which is the best way to add a PERMANENT route add statement so I can route some traffic over one VLAN and some over another?

recently, we needed to configure channel bonding on vlans.
the internet did not supply us with infos, so we did it as a best-effort.

This is a howto for debian etch, but should also work on ubuntu 7.x+:

goal: several vlan-interfaces on top of a bonding trunk

first step: create the bonding config

- in /etc/network/interfaces:

Code:

---

#auto bond0
iface bond0 inet manual
        pre-up ip link set bond0 up
        up ifenslave bond0 eth0
        up ifenslave bond0 eth1
        post-down ifenslave -d bond0 eth0 eth1
        post-down modprobe  -r bond0

---

second step: define the logical network link (repeat as to your needs)

- in /etc/network/interfaces:

Code:

---

auto bond0.911
 allow-hotplug bond0.911
 iface bond0.911 inet static
        address 10.20.50.18
        netmask 255.255.255.192
        network 10.20.50.0
        broadcast 10.20.50.63
        gateway 10.20.50.2
        pre-up /etc/network/bondingstate UP bond0 911
        post-down /etc/network/bondingstate DOWN bond0 911

---

third step: glue them together

- in /etc/network/bondingstate:

Code:

---

# Script /etc/network/bondingstate
# wird aufgerufen aus /etc/network/interfaces
# 04.09.2007, no copyright
#
 
# Variablen setzen

set -x
CHANGE_STATE=$1
TRUNK=$2
VLANID=$3

case $CHANGE_STATE in
  UP)
    #egrep "$TRUNK" /etc/network/run/ifstate >/dev/null 2>&1
    if [ ! -f /proc/net/bonding/$TRUNK ] ; then
      ifup $TRUNK
      if [ $? -ne 0 ] ; then
        exit 1
      fi
    fi
    vconfig add $TRUNK $VLANID
    ;;

  DOWN)
    vconfig rem $TRUNK.$VLANID
    egrep "$TRUNK\.[0-9]+" /etc/network/run/ifstate >/dev/null 2>&1
    if [ $? -ne 0 ] ; then
      if [ -z "$VLANID" ] ; then
        exit 0
      else
        ifdown $TRUNK
      fi
    fi
    ;;
esac

---

What happens:

'auto bond0.911' starts setting up the interface.
'pre-up' calls bondingstate
bondingstate tests whether the bonding module is loaded,
then starts bonding, if need be.
'iface bond0' enslaves the physical NICs, then returns success
bondingstate adds the given vlan ID and returns success
'auto bond0.911' configures IP-addresses and returns success

Configuring additional links is but a copy of 'bond0.911'

On shutdown, 'post-down /etc/network/bondingstate DOWN bond0 911'
calls bondingstate, which in turn checks in /etc/network/run/ifstate
whether this is the last vlan interface and if so, shuts down bond0
bond0's post-down then will remove the kernel module

There might be routing issues. Beware of the order in /etc/network/interfaces.

Hope this might be helpful for anyone, anytime

Have fun,
your swiss telekom hackers
best before end: 09/2007