If you are using the current version of Cumulus Linux, the content on this page may not be up to date. The current version of the documentation is available here. If you are redirected to the main page of the user guide, then this page may have been renamed; please search for it there.

Bonding - Link Aggregation

Linux bonding provides a way to aggregate multiple network interfaces (slaves) into a single logical bonded interface (bond). Link aggregation is useful for linear scaling of bandwidth, load balancing, and failover protection.

Cumulus Linux supports two bonding modes:

  • IEEE 802.3ad link aggregation mode combines one or more links to form a link aggregation group (LAG) so that a media access control (MAC) client can treat the group as a single link. IEEE 802.3ad link aggregation is the default mode.
  • Balance-xor mode balances outgoing traffic across active ports according to the hashed protocol header information and accepts incoming traffic from any active port. All slave interfaces are active for load balancing and fault tolerance. This is useful for MLAG deployments.

Cumulus Linux uses version 1 of the LAG control protocol (LACP).

  • An interface cannot belong to multiple bonds.
  • A bond can have subinterfaces, but subinterfaces cannot have a bond.
  • A bond cannot enslave VLAN subinterfaces.
  • All slave ports within a bond must have the same speed or duplex and match the slave ports of the link partner.

Create a Bond

To create a bond, specify the bond members. In the example below, the front panel port interfaces swp1 thru swp4 are members of bond1 but swp5 and swp6 are not part of bond1.

cumulus@switch:~$ nv set interface bond1 bond member swp1-4
cumulus@switch:~$ nv config apply

In NVUE, if you create the bond interface with a name that starts with bond, NVUE automatically sets the interface type to bond. If you create a bond interface with a name that does not start with bond, you must set the interface type to bond with the nv set interface <interface-name> type bond command.

Edit the /etc/network/interfaces file to add a stanza for the bond, then run the ifreload -a command.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

  • By default, the bond uses IEEE 802.3ad link aggregation mode. To configure the bond in balance-xor mode, see Optional Configuration below.
  • If the bond is not going to be part of a bridge, you must specify an IP address.
  • Make sure the name of the bond adheres to Linux interface naming conventions and is unique within the switch.
  • Cumulus Linux does not support bond members at 200G or greater.
  • To temporarily bring up a bond even when there is no LACP partner, use LACP Bypass.

When you start networking, the switch creates bond1 as MASTER and interfaces swp1 thru swp4 come up in SLAVE mode:

cumulus@switch:~$ ip link show
...

3: swp1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
4: swp2: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
5: swp3: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
6: swp4: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
...

55: bond1: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff

All slave interfaces within a bond have the same MAC address as the bond. Typically, the first slave you add to the bond donates its MAC address as the bond MAC address. The bond MAC address is the source MAC address for all traffic leaving the bond and provides a single destination MAC address to address traffic to the bond.

Removing a bond slave interface from which a bond derives its MAC address affects traffic when the bond interface flaps to update the MAC address.

Optional Configuration

You can set these configuration options for a bond.

Option
Description
Link aggregation modeCumulus Linux supports IEEE 802.3ad link aggregation mode (802.3ad) and balance-xor mode. The default mode is 802.3ad.
Set balance-xor mode only if you cannot use LACP; LACP can detect mismatched link attributes between bond members and can even detect misconnections.

When you use balance-xor mode to dual-connect host-facing bonds in an MLAG environment, you must configure the MLAG ID with the same value on both MLAG switches. Otherwise, the MLAG switch pair treats the bonds as single-connected.

MII link monitoring frequencyHow often (in milliseconds) you want to inspect the link state of each slave for failures.
You can specify a value between 0 and 255. The default value is 100.
miimon link status modeThe miimon link status mode. You can set the mode to either netif_carrier_ok(), or MII or ethtool ioctls. The default setting is netif_carrier_ok().
LACP bypassSet LACP bypass on a bond in 802.3ad mode so that it becomes active and forwards traffic even when there is no LACP partner. You can specify on or off. The default setting is off. See LACP Bypass.
Transmit rateThe rate at which the link partner transmits LACP control packets. You can specify slow or fast. The default setting is fast.
Minimum number of linksThe minimum number of links that must be active before the bond goes into service. You can set a value between 0 and 255. The default value is 1, which indicates that the bond must have at least one active member.

Use a value greater than 1 if you need higher level services to ensure a minimum aggregate bandwidth level before activating a bond.

If the number of active members drops below this setting, the bond appears to upper-level protocols as link-down. When the number of active links returns to greater than or equal to this value, the bond becomes link-up.

Cumulus Linux sets the bond configuration options to the recommended values by default; use caution when changing settings.

To set the link aggregation mode on bond1 to balance-xor mode:

cumulus@switch:~$ nv set interface bond1 bond mode static 
cumulus@switch:~$ nv config apply

To reset the link aggregation mode for bond1 to the default value of 802.3ad, run the nv set interface bond1 bond mode lacp command.

Edit the /etc/network/interfaces file and add the balance-xor parameter to the bond stanza, then run the ifreload -a command:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-mode balance-xor
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

To reset the bond mode for bond1 to the default value of 802.3ad, use the bond-mode 802.3ad parameter:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-mode 802.3ad
    bond-slaves swp1 swp2 swp3 swp4
...

To enable LACP bypass:

cumulus@switch:~$ nv set interface bond1 bond lacp-bypass on 
cumulus@switch:~$ nv config apply

Edit the /etc/network/interfaces file and add the bond-lacp-bypass-allow parameter to the bond stanza, then run the ifreload -a command:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-lacp-bypass-allow
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

To set the miimon link status mode to MII or ethtool ioctls:

Cumulus Linux does not provide NVUE commands for this setting.

Edit the /etc/network/interfaces file and add the bond-use-carrier no parameter to the bond stanza, then run the ifreload -a command:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-use-carrier no
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

To reset the miimon link status mode to the default of netif_carrier_ok(), use the bond-use-carrier yes parameter.

To set the rate at which the link partner transmits LACP control packets to slow:

cumulus@switch:~$ nv set interface bond1 bond lacp-rate slow
cumulus@switch:~$ nv config apply

To reset the rate to the default value of fast, run the nv set interface bond1 bond lacp-rate fast command.

Edit the /etc/network/interfaces file and add the bond-lacp-rate slow parameter to the bond stanza, then run the ifreload -a command:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-lacp-rate slow
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

To reset the rate to the default (fast), use the bond-lacp-rate fast parameter:

To set the minimum number of links that must be active before the bond goes into service to 50:

Cumulus Linux does not provide NVUE commands for this setting.

Edit the /etc/network/interfaces file and add the bond-min-links 50 parameter to the bond stanza, then run the ifreload -a command:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-min-links 50
    bond-slaves swp1 swp2 swp3 swp4
...
cumulus@switch:~$ ifreload -a

Load Balancing

The switch distributes egress traffic through a bond to a slave based on a packet hash calculation, providing load balancing over the slaves; the switch distributes conversation flows over all available slaves to load balance the total traffic. Traffic for a single conversation flow always hashes to the same slave. In a failover event, the switch adjusts the hash calculation to steer traffic over available slaves.

The hash calculation uses packet header data to choose to which slave to transmit the packet:

  • For IP traffic, the switch uses IP header source and destination fields in the calculation.
  • For IP and TCP or UDP traffic, the switch includes source and destination ports in the hash calculation.

For load balancing between multiple interfaces that are members of the same bond, you can hash on these fields:

Field
Default Setting/etc/cumulus/datapath/traffic.conf File Parameter
IP protocolonlag_hash_config.ip_prot
Source MAC addressonlag_hash_config.smac
Destination MAC addressonlag_hash_config.dmac
Source IP addresson
Destination IP addressonlag_hash_config.dip
Source portonlag_hash_config.sport
Destination portonlag_hash_config.dport
Ethertypeonlag_hash_config.ether_type
VLAN IDonlag_hash_config.vlan_id

Cumulus Linux does not provide NVUE commands to configure load balancing.

To set the hash fields:

  1. Edit the /etc/cumulus/datapath/traffic.conf file:

    • Uncomment the lag_hash_config.enable option.
    • Set the hash field to true to include it in the hash calculation or false to omit it from the hash calculation.
  2. Run the echo 1 > /cumulus/switchd/ctrl/hash_config_reload command to reload the configuration. This command does not cause any traffic interruptions.

The following example commands omit the source MAC address and destination MAC address from the hash calculation:

cumulus@switch:~$ sudo nano /etc/cumulus/datapath/traffic.conf
...
#LAG HASH config
#HASH config for LACP to enable custom fields
#Fields will be applicable for LAG hash
#calculation
#Uncomment to enable custom fields configured below
lag_hash_config.enable = true

lag_hash_config.smac = false
lag_hash_config.dmac = false
lag_hash_config.sip  = true
lag_hash_config.dip  = true
lag_hash_config.ether_type = true
lag_hash_config.vlan_id = true
lag_hash_config.sport = true
lag_hash_config.dport = true
lag_hash_config.ip_prot = true
...
cumulus@switch:~$ echo 1 > /cumulus/switchd/ctrl/hash_config_reload

Cumulus Linux enables symmetric hashing by default. Make sure that the settings for the source IP and destination IP fields match, and that the settings for the source port and destination port fields match; otherwise Cumulus Linux disables symmetric hashing automatically. If necessary, you can disable symmetric hashing manually in the /etc/cumulus/datapath/traffic.conf file by setting symmetric_hash_enable = FALSE.

You can also set a unique hash seed for each switch to avoid hash polarization. See Unique Hash Seed.

GTP Hashing

GTP carries mobile data within the core of the mobile operator’s network. Traffic in the 5G Mobility core cluster, from cell sites to compute nodes, have the same source and destination IP address. The only way to identify individual flows is with the GTP TEID. Enabling GTP hashing adds the TEID as a hash parameter and helps the Cumulus Linux switches in the network to distribute mobile data traffic evenly across ECMP routes.

Cumulus Linux supports TEID-based load balancing for traffic egressing a bond and is only applicable if the outer header egressing the port is GTP encapsulated and if the ingress packet is either a GTP-U packet or a VXLAN encapsulated GTP-U packet.

  • GTP Hashing is an early access feature.
  • Cumulus Linux supports GTP Hashing on NVIDIA Spectrum-2 and later.
  • GTP-C packets are not part of GTP hashing.
  • Cumulus Linux does not provide NVUE commands to configure GTP hashing.

To enable TEID-based load balancing:

  1. Edit the /etc/cumulus/datapath/traffic.conf file:

    • Uncomment the hash_config.enable = true line.
    • Change the lag_hash_config.gtp_teid parameter to true.
    cumulus@switch:~$ sudo nano /etc/cumulus/datapath/traffic.conf
    ...
    # Uncomment to enable custom fields configured below
    hash_config.enable = true
    ...
    #GTP-U teid
    lag_hash_config.gtp_teid = true
    
  2. Run the echo 1 > /cumulus/switchd/ctrl/hash_config_reload command. This command does not cause any traffic interruptions.

    cumulus@switch:~$ echo 1 > /cumulus/switchd/ctrl/hash_config_reload
    

To disable TEID-based load balancing, set the lag_hash_config.gtp_teid parameter to false, then reload the configuration.

NVUE does not provide commands to enable TEID-based load balancing.

Troubleshooting

To show information for a bond, run the NVUE nv show interface <bond> bond command:

cumulus@leaf01:mgmt:~$ nv show interface bond1 bond
             operational  applied  description
-----------  -----------  -------  ------------------------------------------------------
down-delay   0            0        bond down delay
lacp-bypass  on           on       lacp bypass
lacp-rate    fast         fast     lacp rate
mode                      lacp     bond mode
up-delay     0            0        bond up delay
[member]     swp1         swp1     Set of bond members
mlag
  enable                  on       Turn the feature 'on' or 'off'.  The default is 'off'.
  id         1            1        MLAG id
  status     single                Mlag Interface status

You can also run the Linux sudo cat /proc/net/bonding/<bond> command:

cumulus@leaf01:mgmt:~$ sudo cat /proc/net/bonding/bond1
...
Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer3+4 (1)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

802.3ad info
LACP rate: fast
Min links: 1
Aggregator selection policy (ad_select): stable
System priority: 65535
System MAC address: 44:38:39:be:ef:aa
Active Aggregator Info:
	Aggregator ID: 1
	Number of ports: 1
	Actor Key: 9
	Partner Key: 1
	Partner Mac Address: 00:00:00:00:00:00

Slave Interface: swp1
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:00:37
Slave queue ID: 0
Aggregator ID: 1
Actor Churn State: none
Partner Churn State: churned
Actor Churned Count: 1
Partner Churned Count: 2
...

To show specific bond information, use the nv show interface <bond> <option> commands:

cumulus@switch:~$ nv show interface bond1 TAB
acl        bridge     ip         lldp       ptp        router     
bond       evpn       link       pluggable  qos
cumulus@leaf02:mgmt:~$ nv show interface bond1 link
                       operational        applied  description
---------------------  -----------------  -------  ----------------------------------------------------------------------
auto-negotiate         off                on       Link speed and characteristic auto negotiation
duplex                 full               full     Link duplex
fec                                       auto     Link forward error correction mechanism
mtu                    9000               9000     interface mtu
speed                  1G                 auto     Link speed
dot1x
  mab                                     off      bypass MAC authentication
  parking-vlan                            off      VLAN for unauthorized MAC addresses
state                  up                 up       The state of the interface
stats
  carrier-transitions  1                           Number of times the interface state has transitioned between up and...
  in-bytes             0 Bytes                     total number of bytes received on the interface
  in-drops             0                           number of received packets dropped
  in-errors            0                           number of received packets with errors
  in-pkts              0                           total number of packets received on the interface
  out-bytes            3.65 MB                     total number of bytes transmitted out of the interface
  out-drops            0                           The number of outbound packets that were chosen to be discarded eve...
  out-errors           0                           The number of outbound packets that could not be transmitted becaus...
  out-pkts             51949                       total number of packets transmitted out of the interface
mac                    44:38:39:00:00:37           MAC Address on an interface