How to configure OVHcloud Link Aggregation in SLES 15 – Support Guides

Learn how to bond your NICs for OVHcloud Link Aggregation in SLES 15.

The OVHcloud Link Aggregation (OLA) technology is designed by our teams to increase your server’s availability and boost the efficiency of your network connections. In just a few clicks, you can aggregate your network cards and make your network links redundant. This means that if one link goes down, the traffic is automatically redirected to another available link.

Requirements

Configuring OVHcloud Link Aggregation in the OVHcloud Control Panel
Access to the OVHcloud Control Panel

Instructions

Because you have a private-private configuration for your NICs in OLA, you will be unable to SSH into the server. Thus, you will need to leverage the IPMI tool to access the server.

To do so, first log in to your OVHcloud Control Panel. In the Bare Metal Cloud section, select your server from Dedicated Servers and click the IPMI tab.

Next, click the From a Java applet (KVM) button.

A JNLP program will be downloaded. Open the program to enter the IPMI. Log in using valid credentials for the server.

By default, using an OVHcloud template, the NICs will be named eth0 and eth1. If you are not using an OVHcloud template, you can find the names of your interfaces using the following command:

ip a

This command will yield numerous "interfaces." If you are having trouble determining which ones are your physical NICs, the first interface will still have the server's public IP address attached to it by default.

Once you have determined the names of your two NICs, you need to configure NIC bonding in the OS. The first step is to create a bond interface. To do so, create the following configuration file in a text editor of your choice:

vi /etc/sysconfig/network/ifcfg-bond0

This will open an empty text file. To configure the bond interface, insert the following into the text file:

STARTMODE='onboot'
BOOTPROTO='static'
IPADDR='10.0.0.1/24'
BONDING_MASTER='yes'
BONDING_SLAVE_0='eth0'
BONDING_SLAVE_1='eth1'
BONDING_MODULE_OPTS='mode=802.3ad xmit_hash_policy=layer3+4'

You can use any private IP address and subnet you wish. If your server has more than 2 network interfaces, you can add them in the configuration, by incrementing the number of the BONDING_SLAVE_ parameter, for example, BONDING_SLAVE_2='eth2'.

Save and exit the file once you have confirmed that the information is correct. Next, you need to configure both physical interfaces. By default, on an OVHcloud server, only eth0 will have a configuration file. Open it using the following command:

vi /etc/sysconfig/network/ifcfg-eth0

By default, the file will appear as follows:

BOOTPROTO=dhcp4
IPADDR6=2001:41d0:408:dd00::/56
LLADDR=10:70:fd:c5:14:00
STARTMODE=auto

NOTE: The IP addresses will be different for each server.

Edit the file to make it appear as follows:

BOOTPROTO='none'
#IPADDR6=2001:41d0:408:dd00::/56
LLADDR=10:70:fd:c5:14:00
STARTMODE='hotplug'

The hardware address (MAC address) of the NIC can be found using the ip a command that you used earlier. It will be the number next to link/ether in the output.

The # in front of a line means that the server will ignore this line when reading the file. Thus, please ignore these lines entirely when creating your interface file for eth0.

Create the eth1 configuration file using the following command:

vi /etc/sysconfig/network-scripts/ifcfg-eth2

This time, the file will be blank, so add the following content to the file:

BOOTPROTO='none'
STARTMODE='hotplug'
LLADDR=0c:42:a1:a7:29:c2

Finally, restart the network daemon using the following command:

wicked ifreload all

To test that your bond is working, ping another server on the same vRack. If it works, you are all set. If it does not, double-check your configurations or try rebooting the server.

You can also check the used parameters by your ifcfg-bond0 interface using the following command: