Setting up an MC-LAG aggregation group

Written by

Updated at December 1, 2025

Servers with MC-LAG support use two network adapters to simultaneously connect to each of the networks (public and private). To ensure fault tolerance, each of the network interface pairs connected to the networks must form an aggregation group on the server side. For more information, see Configuring aggregation groups and network interfaces.

Note

When setting up aggregation groups, do not connect to the server via the network interface you are going to include in a group: the connection with be lost when you create the group. The KVM console is the most reliable way to configure MC-LAG groups.

Currently, you can set up MC-LAG groups in Linux Ubuntu 20.04, 22.04, 24.04, and Debian 11. As an example, this guide uses a server with two pairs of network adapters with connection speed of 25 Gbps each.

To set up a link aggregation group:

Ubuntu/Debian (Netplan)

Install ethtool:
```
apt install ethtool
```

Make sure the required network interfaces are installed in the system and active:

ip link

Result:

...
2: etx3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d6 brd ff:ff:ff:ff:ff:ff
3: etx4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d7 brd ff:ff:ff:ff:ff:ff
4: etx1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether 58:a2:e1:ad:38:2a brd ff:ff:ff:ff:ff:ff
5: etx2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether 58:a2:e1:ad:38:2b brd ff:ff:ff:ff:ff:ff

As you can see from the output, the server has four active network interfaces:

etx3: With the b8:ce:f6:40:12:d6 MAC address.
etx4: With the b8:ce:f6:40:12:d7 MAC address.
etx1: With the 58:a2:e1:ad:38:2a MAC address.
etx2: With the 58:a2:e1:ad:38:2b MAC address.

Find out which of the interfaces belong to the public network, and which to the private one:
1. In the management console, select the folder the server belongs to.
2. Go to BareMetal and select the server in the list of servers.
  
  On the page that opens, in the MAC address field under Public network and Private network, you can see the MAC addresses of interfaces connected to the public and private networks, respectively.
3. Use the information obtained in the two previous steps to identify the server interface pairs connected to the public and private networks. In the example above, the pairs are as follows:
  
  Public network:
  - etx3: With the b8:ce:f6:40:12:d6 MAC address.
  - etx1: With the 58:a2:e1:ad:38:2a MAC address.
  Private network:
  - etx4: With the b8:ce:f6:40:12:d7 MAC address.
  - etx2: With the 58:a2:e1:ad:38:2b MAC address.
Find out the name of the Netplan configuration file:
```
ls /etc/netplan/
```
Result:
```
50-cloud-init.yaml
```
Open the Netplan configuration file with a text editor. In this guide, we use nano:
```
nano /etc/netplan/50-cloud-init.yaml
```

Edit the Netplan configuration by adding aggregation groups (the bonds section):

network:
    bonds:
        bond1:
            dhcp4: true
            interfaces:
            - <public_interface_1_name>
            - <public_interface_2_name>
            macaddress: <public_interface_1_or_2_MAC_address>
            parameters:
                lacp-rate: fast
                mode: 802.3ad
                transmit-hash-policy: layer3+4
        bond2:
            dhcp4: true
            interfaces:
            - <private_interface_1_name>
            - <private_interface_2_name>
            macaddress: <private_interface_1_or_2_MAC_address>
            parameters:
                lacp-rate: fast
                mode: 802.3ad
                transmit-hash-policy: layer3+4
    ethernets:
        etx1:
            dhcp4: false
            match:
                macaddress: 58:a2:e1:ad:38:2a
            set-name: etx1
        etx2:
            dhcp4: false
            match:
                macaddress: 58:a2:e1:ad:38:2b
            set-name: etx2
        etx3:
            dhcp4: false
            match:
                macaddress: b8:ce:f6:40:12:d6
            set-name: etx3
        etx4:
            dhcp4: false
            match:
                macaddress: b8:ce:f6:40:12:d7
            set-name: etx4
    version: 2

Where:

<public_interface_1_name>, <public_interface_2_name>: Names of the interfaces which belong to the public network, as you found out earlier.
<public_interface_1_or_2_MAC_address>: MAC address of an interface which belongs to the public network, as you found out earlier.
<private_interface_1_name>, <public_interface_2_name>: Names of the interfaces which belong to the private network, as you found out earlier.
<private_interface_1_or_2_MAC_address>: MAC address of an interface which belongs to the private network, as you found out earlier.

Warning

Note that DHCP must be:

Enabled (dhcp4: true) for aggregation groups (the bonds section).
Disabled (dhcp4: false) for individual interfaces (the ethernets section).

Netplan configuration example

network:
    bonds:
        bond1:
            dhcp4: true
            interfaces:
            - etx3
            - etx1
            macaddress: b8:ce:f6:40:12:d6
            parameters:
                lacp-rate: fast
                mode: 802.3ad
                transmit-hash-policy: layer3+4
        bond2:
            dhcp4: true
            interfaces:
            - etx4
            - etx2
            macaddress: b8:ce:f6:40:12:d7
            parameters:
                lacp-rate: fast
                mode: 802.3ad
                transmit-hash-policy: layer3+4
    ethernets:
        etx1:
            dhcp4: false
            match:
                macaddress: 58:a2:e1:ad:38:2a
            set-name: etx1
        etx2:
            dhcp4: false
            match:
                macaddress: 58:a2:e1:ad:38:2b
            set-name: etx2
        etx3:
            dhcp4: false
            match:
                macaddress: b8:ce:f6:40:12:d6
            set-name: etx3
        etx4:
            dhcp4: false
            match:
                macaddress: b8:ce:f6:40:12:d7
            set-name: etx4
    version: 2

Apply the new Netplan configuration:
```
netplan apply
```

Make sure the list of network interfaces displays the aggregation groups:

ip link

Result:

...
2: etx3: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq master bond1 state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d6 brd ff:ff:ff:ff:ff:ff
3: etx4: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq master bond2 state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d7 brd ff:ff:ff:ff:ff:ff
4: etx1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq master bond1 state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d6 brd ff:ff:ff:ff:ff:ff
5: etx2: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq master bond2 state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d7 brd ff:ff:ff:ff:ff:ff
6: bond1: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d6 brd ff:ff:ff:ff:ff:ff
7: bond2: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
    link/ether b8:ce:f6:40:12:d7 brd ff:ff:ff:ff:ff:ff

As you can see from the output, the server now has two MC-LAG aggregation groups, bond1 and bond2.

Note

If aggregation groups are inactive (DOWN), activate them:

ip link set bond1 up
ip link set bond2 up

View the information about the groups you created. As an example, let’s use the aggregation group connected to the public network:

ethtool bond1

Result:

Settings for bond1:
  Supported ports: [ ]
  Supported link modes:   Not reported
  Supported pause frame use: No
  Supports auto-negotiation: No
  Supported FEC modes: Not reported
  Advertised link modes:  Not reported
  Advertised pause frame use: No
  Advertised auto-negotiation: No
  Advertised FEC modes: Not reported
  Speed: 50000Mb/s
  Duplex: Full
  Port: Other
  PHYAD: 0
  Transceiver: internal
  Auto-negotiation: off
  Link detected: yes

As you can see from the output, the connection speed for the bond1 group is 50 Gbps.

Simulate an incident where a link in the bond1 public network aggregation group fails. To do this, disable one of the group’s network interfaces:
```
ip link set etx3 down
```

View the group information again:

ethtool bond1

Result:

Settings for bond1:
  Supported ports: [ ]
  Supported link modes:   Not reported
  Supported pause frame use: No
  Supports auto-negotiation: No
  Supported FEC modes: Not reported
  Advertised link modes:  Not reported
  Advertised pause frame use: No
  Advertised auto-negotiation: No
  Advertised FEC modes: Not reported
  Speed: 25000Mb/s
  Duplex: Full
  Port: Other
  PHYAD: 0
  Transceiver: internal
  Auto-negotiation: off
  Link detected: yes

As you can see from the output, the connection speed for the bond1 group has been reduced to 25 Gbps, but the network connectivity has been maintained. To check this, connect to the server over SSH.

Activate the interface you disabled earlier and make sure the aggregation group is running again at the maximum speed:

ip link set etx3 up
ethtool bond1

Result:

Settings for bond1:
  Supported ports: [ ]
  Supported link modes:   Not reported
  Supported pause frame use: No
  Supports auto-negotiation: No
  Supported FEC modes: Not reported
  Advertised link modes:  Not reported
  Advertised pause frame use: No
  Advertised auto-negotiation: No
  Advertised FEC modes: Not reported
  Speed: 50000Mb/s
  Duplex: Full
  Port: Other
  PHYAD: 0
  Transceiver: internal
  Auto-negotiation: off
  Link detected: yes

Setting up an MC-LAG aggregation group

See also

Was the article helpful?

Setting up an MC-LAG aggregation group

See alsoSee also

Was the article helpful?

See also