|
Enabled
|
Enables a bond
|
|
Name
|
Bond name
|
|
Node name
|
A node in the UserGate cluster where a new bond will be created
|
|
Zone
|
A zone to which the bond will belong
|
|
Interfaces
|
One or more interfaces that will be used for creating the bond
|
|
Mode
|
The working mode of the bond must be the same as that of the device to which the bond will be connected. Possible options:
-
Round robin. Packets are sent one by one, starting with the first available interface and ending with the last available interface. This policy is used for better load balancing and failover.
-
Active backup. Only one network interface among the merged interfaces will be active. Another interface will be activated only when the currently active interface goes down. With such a policy, MAC address of a bond interface can be obtained from the outside using only one network port in order to avoid switching issues. This policy is used for failover.
-
XOR. The traffic is distributed among network adapters using the following formula: [("MAC address of the origin" XOR "MAC address of the destination") modulo "number of interfaces"]. This means that the same network adapter sends packets to the same destinations. Optionally, the traffic distribution can also be based on the xmit_hash policy. This XOR policy is used for load balancing and failover.
-
Broadcast. Sends all packets to all network interfaces. This policy is used for failover.
-
IEEE 802.3ad - this is a default working mode supported by most network switches. The system creates aggregated groups of network adapters with the same throughput and duplex. In this mode, packets are sent through all channels in the active aggregate according to IEEE 802.3ad. Which interface to use for sending packets depends on the current policy: by default, it is XOR policy, but the xmit_hash policy is also supported.
-
Adaptive transmit load balancing. The outgoing traffic is distributed depending on load on each network adapter (i.e. download speed) and does not require any additional configuration on the switch. The incoming traffic goes to the current network adapter. In case of its failure, another network adapter will be assigned the MAC address of the failed one.
-
Adaptive load balancing. Includes the above-mentioned policy plus load balancing for the incoming traffic. This option does not require any additional configuration on the switch. The incoming traffic is balanced through ARP communications. The driver captures ARP responses sent from local network adapters to the outside, and then replaces the MAC address of the origin with one of the unique MAC addresses of the network adapters to be bound. As a result, different peers use different MAC addresses of the server. Balancing of the incoming traffic is distributed among interfaces one by one (round-robin).
|
|
MII monitoring period (ms)
|
Sets the frequency of MII monitoring (in milliseconds). Sets the frequency of checks for failures in a communication line. The default value is 0 - disable MII monitoring.
|
|
Down delay (ms)
|
Sets a delay (in milliseconds) before the interface goes down due to a connection failure. This option is valid only when MII monitoring (miimon) is enabled. Values of this parameter must be divisible by "miimon" values. When a value is not divisible, it will be rounded to the nearest divisible value. The default value is 0.
|
|
Up delay (ms)
|
Sets a delay (in milliseconds) before the interface goes up after connection recovery. This parameter is valid only when MII monitoring (miimon) is enabled. Values of this parameter must be divisible by "miimon" values. When a value is not divisible, it will be rounded to the nearest divisible value. The default value is 0.
|
|
LACP rate
|
Sets the interval of sending LACPDU packets by a partner in the 802.3ad mode. Possible values:
|
|
Failover MAC
|
Sets how to assign MAC addresses to the merged interfaces in the active-backup mode when switching between the interfaces. The standard behavior is to assign the same MAC address to all interfaces. Possible values:
-
Disabled - the same MAC address is assigned to all interfaces when switching between them
-
Active - the MAC address of the bond interface will always be the same as that of the currently active interface. MAC addresses of the backup interfaces are not changed. The MAC address of the bond interface is changed when handling a failure.
-
Follow - the MAC address of the bond interface will be the same as that of the first interface added to the bond. This MAC address will not be assigned to the second interface and so on while they are in the backup mode. The MAC address is assigned when handling a failure: the backup interface becomes active, obtains a new MAC (the same as of the bond interface) while the previously active interface obtains the MAC that was used by the currently active interface.
|
|
Xmit hash policy
|
Sets the hash policy of sending packets through the merged interfaces in the XOR or IEEE 802.3ad modes. Possible values:
-
Layer 2 - only MAC addresses are used for generating a hash. When this option is enabled, traffic for a certain network host will always be sent through the same interface in accordance with IEEE 802.3ad.
-
Layer 2+3 - both MAC addresses and IP addresses are used for generating a hash in accordance with IEEE 802.3ad.
-
Layer 3+4 - IP addresses and transport layer protocols (TCP or UDP) are used for generating a hash. This option is not fully compatible with IEEE 802.3ad, since both fragmented and non-fragmented packets can be sent within the same TCP or UDP communication. Fragmented packets do not contain origin ports and destination ports. As a result, packets within the same session can be obtained at the destination point in a different order, since they are sent through different interfaces.
|
|
Network
|
Assignment of IP addresses: no address, static IP address or dynamic IP address obtained through DHCP.
|
|
DHCP relay
|
Configuring a DHCP relay for a bond interface. Enable a DHCP relay, then in the UserGate address field, enter the IP address of the interface to which you want to add a relay, and specify one or more DHCP servers to which DHCP queries from clients should be routed.
|
