4.5. Using Channel Bonding

To enhance performance, adjust available module options to ascertain what combination works best. Pay particular attention to the miimon or arp_interval and the arp_ip_target parameters. See Section 4.5.1, “Bonding Module Directives” for a list of available options and how to quickly determine the best ones for your bonded interface.

4.5.1. Bonding Module Directives

It is a good idea to test which channel bonding module parameters work best for your bonded interfaces before adding them to the BONDING_OPTS="bonding parameters" directive in your bonding interface configuration file (ifcfg-bond0 for example). Parameters to bonded interfaces can be configured without unloading (and reloading) the bonding module by manipulating files in the sysfs file system.
sysfs is a virtual file system that represents kernel objects as directories, files and symbolic links. sysfs can be used to query for information about kernel objects, and can also manipulate those objects through the use of normal file system commands. The sysfs virtual file system is mounted under the /sys/ directory. All bonding interfaces can be configured dynamically by interacting with and manipulating files under the /sys/class/net/ directory.
In order to determine the best parameters for your bonding interface, create a channel bonding interface file such as ifcfg-bond0 by following the instructions in Section 4.4.2, “Create a Channel Bonding Interface”. Insert the SLAVE=yes and MASTER=bond0 directives in the configuration files for each interface bonded to bond0. Once this is completed, you can proceed to testing the parameters.
First, open the bond you created by running ifup bondN as root:
~]# ifup bond0
If you have correctly created the ifcfg-bond0 bonding interface file, you will be able to see bond0 listed in the output of running ip link show as root:
~]# ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 1000
    link/ether 52:54:00:e9:ce:d2 brd ff:ff:ff:ff:ff:ff
3: eth1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 1000
    link/ether 52:54:00:38:a6:4c brd ff:ff:ff:ff:ff:ff
4: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT
    link/ether 52:54:00:38:a6:4c brd ff:ff:ff:ff:ff:ff
To view all existing bonds, even if they are not up, run:
~]$ cat /sys/class/net/bonding_masters
bond0
You can configure each bond individually by manipulating the files located in the /sys/class/net/bondN/bonding/ directory. First, the bond you are configuring must be taken down:
~]# ifdown bond0
As an example, to enable MII monitoring on bond0 with a 1 second interval, run as root:
~]# echo 1000 > /sys/class/net/bond0/bonding/miimon
To configure bond0 for balance-alb mode, run either:
~]# echo 6 > /sys/class/net/bond0/bonding/mode
...or, using the name of the mode:
~]# echo balance-alb > /sys/class/net/bond0/bonding/mode
After configuring options for the bond in question, you can bring it up and test it by running ifup bondN. If you decide to change the options, take the interface down, modify its parameters using sysfs, bring it back up, and re-test.
Once you have determined the best set of parameters for your bond, add those parameters as a space-separated list to the BONDING_OPTS= directive of the /etc/sysconfig/network-scripts/ifcfg-bondN file for the bonding interface you are configuring. Whenever that bond is brought up (for example, by the system during the boot sequence if the ONBOOT=yes directive is set), the bonding options specified in the BONDING_OPTS will take effect for that bond.
The following list provides the names of many of the more common channel bonding parameters, along with a description of what they do. For more information, see the brief descriptions for each parm in modinfo bonding output, or for more detailed information, see https://www.kernel.org/doc/Documentation/networking/bonding.txt.

Bonding Interface Parameters

ad_select=value
Specifies the 802.3ad aggregation selection logic to use. Possible values are:
  • stable or 0 — Default setting. The active aggregator is chosen by largest aggregate bandwidth. Reselection of the active aggregator occurs only when all slaves of the active aggregator are down or if the active aggregator has no slaves.
  • bandwidth or 1 — The active aggregator is chosen by largest aggregate bandwidth. Reselection occurs if:
    • A slave is added to or removed from the bond;
    • Any slave's link state changes;
    • Any slave's 802.3ad association state changes;
    • The bond's administrative state changes to up.
  • count or 2 — The active aggregator is chosen by the largest number of slaves. Reselection occurs as described for the bandwidth setting above.
The bandwidth and count selection policies permit failover of 802.3ad aggregations when partial failure of the active aggregator occurs. This keeps the aggregator with the highest availability, either in bandwidth or in number of slaves, active at all times.
arp_interval=time_in_milliseconds
Specifies, in milliseconds, how often ARP monitoring occurs.

Important

It is essential that both arp_interval and arp_ip_target parameters are specified, or, alternatively, the miimon parameter is specified. Failure to do so can cause degradation of network performance in the event that a link fails.
If using this setting while in mode=0 or mode=2 (the two load-balancing modes), the network switch must be configured to distribute packets evenly across the NICs. For more information on how to accomplish this, see https://www.kernel.org/doc/Documentation/networking/bonding.txt.
The value is set to 0 by default, which disables it.
arp_ip_target=ip_address[,ip_address_2,…ip_address_16]
Specifies the target IP address of ARP requests when the arp_interval parameter is enabled. Up to 16 IP addresses can be specified in a comma separated list.
arp_validate=value
Validate source/distribution of ARP probes; default is none. Other valid values are active, backup, and all.
downdelay=time_in_milliseconds
Specifies (in milliseconds) how long to wait after link failure before disabling the link. The value must be a multiple of the value specified in the miimon parameter. The value is set to 0 by default, which disables it.
fail_over_mac=value
Specifies whether active-backup mode should set all slaves to the same MAC address at enslavement (the traditional behavior), or, when enabled, perform special handling of the bond's MAC address in accordance with the selected policy. Possible values are:
  • none or 0 — Default setting. This setting disables fail_over_mac, and causes bonding to set all slaves of an active-backup bond to the same MAC address at enslavement time.
  • active or 1 — The active fail_over_mac policy indicates that the MAC address of the bond should always be the MAC address of the currently active slave. The MAC address of the slaves is not changed; instead, the MAC address of the bond changes during a failover.
    This policy is useful for devices that cannot ever alter their MAC address, or for devices that refuse incoming broadcasts with their own source MAC (which interferes with the ARP monitor). The disadvantage of this policy is that every device on the network must be updated by gratuitous ARP, as opposed to the normal method of switches snooping incoming traffic to update their ARP tables. If the gratuitous ARP is lost, communication may be disrupted.
    When this policy is used in conjunction with the MII monitor, devices which assert link up prior to being able to actually transmit and receive are particularly susceptible to loss of the gratuitous ARP, and an appropriate updelay setting may be required.
  • follow or 2 — The follow fail_over_mac policy causes the MAC address of the bond to be selected normally (normally the MAC address of the first slave added to the bond). However, the second and subsequent slaves are not set to this MAC address while they are in a backup role; a slave is programmed with the bond's MAC address at failover time (and the formerly active slave receives the newly active slave's MAC address).
    This policy is useful for multiport devices that either become confused or incur a performance penalty when multiple ports are programmed with the same MAC address.
lacp_rate=value
Specifies the rate at which link partners should transmit LACPDU packets in 802.3ad mode. Possible values are:
  • slow or 0 — Default setting. This specifies that partners should transmit LACPDUs every 30 seconds.
  • fast or 1 — Specifies that partners should transmit LACPDUs every 1 second.
miimon=time_in_milliseconds
Specifies (in milliseconds) how often MII link monitoring occurs. This is useful if high availability is required because MII is used to verify that the NIC is active. To verify that the driver for a particular NIC supports the MII tool, type the following command as root:
~]# ethtool interface_name | grep "Link detected:"
In this command, replace interface_name> with the name of the device interface, such as eth0, not the bond interface. If MII is supported, the command returns:
Link detected: yes
If using a bonded interface for high availability, the module for each NIC must support MII. Setting the value to 0 (the default), turns this feature off. When configuring this setting, a good starting point for this parameter is 100.

Important

It is essential that both arp_interval and arp_ip_target parameters are specified, or, alternatively, the miimon parameter is specified. Failure to do so can cause degradation of network performance in the event that a link fails.
mode=value
Allows you to specify the bonding policy. The value can be one of:
  • balance-rr or 0 — Sets a round-robin policy for fault tolerance and load balancing. Transmissions are received and sent out sequentially on each bonded slave interface beginning with the first one available.
  • active-backup or 1 — Sets an active-backup policy for fault tolerance. Transmissions are received and sent out through the first available bonded slave interface. Another bonded slave interface is only used if the active bonded slave interface fails.
  • balance-xor or 2 — Transmissions are based on the selected hash policy. The default is to derive a hash by XOR of the source and destination MAC addresses multiplied by the modulo of the number of slave interfaces. In this mode traffic destined for specific peers will always be sent over the same interface. As the destination is determined by the MAC addresses this method works best for traffic to peers on the same link or local network. If traffic has to pass through a single router then this mode of traffic balancing will be suboptimal.
  • broadcast or 3 — Sets a broadcast policy for fault tolerance. All transmissions are sent on all slave interfaces.
  • 802.3ad or 4 — Sets an IEEE 802.3ad dynamic link aggregation policy. Creates aggregation groups that share the same speed and duplex settings. Transmits and receives on all slaves in the active aggregator. Requires a switch that is 802.3ad compliant.
  • balance-tlb or 5 — Sets a Transmit Load Balancing (TLB) policy for fault tolerance and load balancing. The outgoing traffic is distributed according to the current load on each slave interface. Incoming traffic is received by the current slave. If the receiving slave fails, another slave takes over the MAC address of the failed slave. This mode is only suitable for local addresses known to the kernel bonding module and therefore cannot be used behind a bridge with virtual machines.
  • balance-alb or 6 — Sets an Adaptive Load Balancing (ALB) policy for fault tolerance and load balancing. Includes transmit and receive load balancing for IPv4 traffic. Receive load balancing is achieved through ARP negotiation. This mode is only suitable for local addresses known to the kernel bonding module and therefore cannot be used behind a bridge with virtual machines.
primary=interface_name
Specifies the interface name, such as eth0, of the primary device. The primary device is the first of the bonding interfaces to be used and is not abandoned unless it fails. This setting is particularly useful when one NIC in the bonding interface is faster and, therefore, able to handle a bigger load.
This setting is only valid when the bonding interface is in active-backup mode. See https://www.kernel.org/doc/Documentation/networking/bonding.txt for more information.
primary_reselect=value
Specifies the reselection policy for the primary slave. This affects how the primary slave is chosen to become the active slave when failure of the active slave or recovery of the primary slave occurs. This parameter is designed to prevent flip-flopping between the primary slave and other slaves. Possible values are:
  • always or 0 (default) — The primary slave becomes the active slave whenever it comes back up.
  • better or 1 — The primary slave becomes the active slave when it comes back up, if the speed and duplex of the primary slave is better than the speed and duplex of the current active slave.
  • failure or 2 — The primary slave becomes the active slave only if the current active slave fails and the primary slave is up.
The primary_reselect setting is ignored in two cases:
  • If no slaves are active, the first slave to recover is made the active slave.
  • When initially enslaved, the primary slave is always made the active slave.
Changing the primary_reselect policy through sysfs will cause an immediate selection of the best active slave according to the new policy. This may or may not result in a change of the active slave, depending upon the circumstances
resend_igmp=range
Specifies the number of IGMP membership reports to be issued after a failover event. One membership report is issued immediately after the failover, subsequent packets are sent in each 200ms interval.
The valid range is 0 to 255; the default value is 1. A value of 0 prevents the IGMP membership report from being issued in response to the failover event.
This option is useful for bonding modes balance-rr (mode 0), active-backup (mode 1), balance-tlb (mode 5) and balance-alb (mode 6), in which a failover can switch the IGMP traffic from one slave to another. Therefore a fresh IGMP report must be issued to cause the switch to forward the incoming IGMP traffic over the newly selected slave.
updelay=time_in_milliseconds
Specifies (in milliseconds) how long to wait before enabling a link. The value must be a multiple of the value specified in the miimon parameter. The value is set to 0 by default, which disables it.
use_carrier=number
Specifies whether or not miimon should use MII/ETHTOOL ioctls or netif_carrier_ok() to determine the link state. The netif_carrier_ok() function relies on the device driver to maintains its state with netif_carrier_on/off; most device drivers support this function.
The MII/ETHTOOL ioctls tools utilize a deprecated calling sequence within the kernel. However, this is still configurable in case your device driver does not support netif_carrier_on/off.
Valid values are:
  • 1 — Default setting. Enables the use of netif_carrier_ok().
  • 0 — Enables the use of MII/ETHTOOL ioctls.

Note

If the bonding interface insists that the link is up when it should not be, it is possible that your network device driver does not support netif_carrier_on/off.
xmit_hash_policy=value
Selects the transmit hash policy used for slave selection in balance-xor and 802.3ad modes. Possible values are:
  • 0 or layer2 — Default setting. This parameter uses the XOR of hardware MAC addresses to generate the hash. The formula used is:
    (source_MAC_address XOR destination_MAC) MODULO slave_count
    This algorithm will place all traffic to a particular network peer on the same slave, and is 802.3ad compliant.
  • 1 or layer3+4 — Uses upper layer protocol information (when available) to generate the hash. This allows for traffic to a particular network peer to span multiple slaves, although a single connection will not span multiple slaves.
    The formula for unfragmented TCP and UDP packets used is:
    ((source_port XOR dest_port) XOR
      ((source_IP XOR dest_IP) AND 0xffff)
        MODULO slave_count
    For fragmented TCP or UDP packets and all other IP protocol traffic, the source and destination port information is omitted. For non-IP traffic, the formula is the same as the layer2 transmit hash policy.
    This policy intends to mimic the behavior of certain switches; particularly, Cisco switches with PFC2 as well as some Foundry and IBM products.
    The algorithm used by this policy is not 802.3ad compliant.
  • 2 or layer2+3 — Uses a combination of layer2 and layer3 protocol information to generate the hash.
    Uses XOR of hardware MAC addresses and IP addresses to generate the hash. The formula is:
    (((source_IP XOR dest_IP) AND 0xffff) XOR
      ( source_MAC XOR destination_MAC ))
        MODULO slave_count
    This algorithm will place all traffic to a particular network peer on the same slave. For non-IP traffic, the formula is the same as for the layer2 transmit hash policy.
    This policy is intended to provide a more balanced distribution of traffic than layer2 alone, especially in environments where a layer3 gateway device is required to reach most destinations.
    This algorithm is 802.3ad compliant.