LACP

You can run the Link Aggregation Control Protocol (LACP) on Extreme Networks devices. LACP enables dynamic load sharing and hot standby for link aggregation links, in accordance with the IEEE 802.3ad standard. All third-party devices supporting LACP run with Extreme Networks devices.

The addition of LACP provides the following enhancements to static load sharing, or link aggregation:

After you enable load-sharing, the LACP protocol is enabled by default. You configure dynamic link aggregation by first assigning a primary, or logical, port to the group, or LAG and then specifying the other ports you want in the LAG.

LACP, using an automatically generated key, determines which links can aggregate. Each link can belong to only one LAG. LACP determines which links are available. The communicating systems negotiate priority for controlling the actions of the entire trunk (LAG), using LACP, based on the lowest system MAC number. You can override this automatic prioritization by configuring the system priority for each LAG.

After you enable and configure LACP, the system sends PDUs (LACPDUs) on the LAG ports. The LACPDUs inform the remote system of the identity of the sending system, the automatically generated key of the link, and the desired aggregation capabilities of the link. If a key from a particular system on a given link matches a key from that system on another link, those links are aggregatable. After the remote system exchanges LACPDUs with the LAG, the system determines the status of the ports and whether to send traffic on which ports.

Among those ports deemed aggregatable by LACP, the system uses those ports with the lowest port number as active ports; the remaining ports aggregatable to that LAG are put into standby status. Should an active link fail, the standby ports become active, also according to the lowest port number. (See Configuring LACP for the number of active and standby LACP links supported per platform.)

All ports configured in a LAG begin in an unselected state. Based on the LACPDUs exchanged with the remote link, those ports that have a matching key are moved into a selected state. If there is no matching key, the ports in the LAG remain in the unselected state.

However,if more ports in the LAG are selected than the aggregator can handle because of the system hardware, those ports that fall out of the hardware‘s capability are moved into standby state. The lowest numbered ports are the first to be automatically added to the aggregator; the rest go to standby. As the name implies, these ports are available to join the aggregator if one of the selected ports should go offline.

You can configure the port priority to ensure the order that ports join the aggregator. However, that port must first be added to the LAG before you can configure the LACP settings. Again, if more than one port is configured with the same priority, the lowest-numbered port joins the aggregator first.

After the ports in the LAG move into the selected state, LACP uses the mux portion of the protocol to determine which ports join the aggregator and can collect and distribute traffic. A few seconds after a port is selected, it moves into the mux state of waiting, and then into the mux state of attached. The attached ports then send their own LACP sync messages announcing that they are ready to receive traffic.

The protocol keeps sending and receiving LACPDUs until both sides of the link have echoed back each other‘s information; the ends of the link are then considered synchronized. After the sync messages match up on each end, that port is moved into the aggregator (into the mux state of collecting-distributing) and is able to collect and distribute traffic.

The protocol then enables the aggregated link for traffic and monitors the status of the links for changes that may require reconfiguration. For example, if one of the links in a LAG goes down and there are standby links in that LAG, LACP automatically moves the standby port into selected mode and that port begins collecting and distributing traffic.

The marker protocol portion of LACP ensures that all traffic on a link has been received in the order in which it was sent and is used when links must be dynamically moved between aggregation groups. The Extreme Networks LACP implementation responds to marker frames but does not initiate these frames.

Note

Note

Always verify the LACP configuration by issuing the show ports sharing command; look for the ports specified as being in the aggregator. You can also display the aggregator count by issuing the show lacp lag command.

You can configure additional parameters for the LACP protocol and the system sends certain SNMP traps in conjunction with LACP. The system sends a trap when a member port is added to or deleted from an aggregator.

The system now detects and blocks loopbacks; that is, the system does not allow a pair of ports that are in the same LAG but are connected to one another by the same link to select the same aggregator. If a loopback condition exists between two ports, they cannot aggregate. Ports with the same MAC address and the same admin key cannot aggregate; ports with the same MAC address and a different admin key can belong to the same LAG.

The system sends an error message if a LAG port is configured and up but still not attached to the aggregator or in operation within 60 seconds. Use the show lacp member-port port detail command to display the churn on both sides of the link. If the churn value is shown as True in the display, check your LACP configuration. The issue may be either on your end or on the partner link, but you should check your configuration. The display shows as True until the aggregator forms, and then it changes to False.

A LAG port moves to expired and then to the defaulted state when it fails to receive an LACPDU from its partner for a specified time. You can configure this timeout value as long, which is 90 seconds, or short, which is three seconds; the default is long. (In ExtremeXOS 11.3, the timeout value is not configurable and is set as long, or 90 seconds.) Use the show lacp lag group-id detail command to display the timeout value for the LAG.

There are two LACP activity modes: active and passive. In LACP active mode, the switch periodically sends LACPDUs; in passive mode, the switch sends LACPDUs only when it receives one from the other end of the link. The default is active mode. (In ExtremeXOS 11.3, the mode is not configurable; it is always active mode.) Use the show lacp lag group-id detail command to display the LACP mode for the LAG.

Note

Note

One side of the link must be in active mode in order to pass traffic. If you configure your side in the passive mode, ensure that the partner link is in LACP active mode.
A LAG port moves into a defaulted state after the timeout value expires with no LACPDUs received for the other side of the link. You can configure whether you want this defaulted LAG port removed from the aggregator or added back into the aggregator. If you configure the LAG to remove ports that move into the default state, those ports are removed from the aggregator and the port state is set to Unselected. The default configuration for defaulted ports is to be removed, or deleted, from the aggregator. (In ExtremeXOS version 11.3, defaulted ports in the LAG are always removed from the aggregator; this is not configurable.)
Note

Note

To force the LACP trunk to behave like a static sharing trunk, use the configure sharing port lacp defaulted-state-action [add | delete] command to add ports to the aggregator.

If you configure the LAG to add the defaulted port into the aggregator, the system takes inventory of the number of ports currently in the aggregator. If there are fewer ports in the aggregator than the maximum number allowed, the system adds the defaulted port to the aggregator (port set to selected and collecting-distributing). If the aggregator has the maximum ports, the system adds the defaulted port to the standby list (port set to standby). Use the show lacp lag group-id {detail} command to display the defaulted action set for the LAG.

Note

Note

If the defaulted port is assigned to standby, that port automatically has a lower priority than any other port in the LAG (including those already in standby).