Dell EMC SmartFabric OS10 User Guide Release 10.5.0

Graceful LACP with VLT

When a VLT node is reloaded, all its interfaces including VLT port channel interfaces go down. Top-of-Rack (ToR) devices that are connected at the other end of the VLT port channel interfaces could take a considerable amount of time to detect the interface status change and switch the traffic towards the other active VLT node. Using LACP PDUs, the graceful LACP feature enables VLT nodes to inform ToR devices ahead of taking down the member ports of its VLT port channel interfaces. Thus, the graceful LACP feature enables the ToR devices to switch the traffic to the other active VLT node.

Graceful LACP is supported in these scenarios:

• When a VLT node is reloaded
• When the secondary VLT node detects that the VLTi link is down but the heartbeat is functional

Graceful LACP is enabled by default and you cannot disable it.

The following shows the normal behavior of a VLT setup where data flows through the optimal path to its destination:

The following shows a scenario where VLT Peer A is being reloaded or going down:

Until LACP convergence happens, the server continues to forward traffic to VLT Peer A resulting in traffic loss for a longer time interval.

With graceful LACP, VLT Peer A sends graceful LACP PDUs out to all VLT member ports, as shown:

These PDUs notify the server to direct the traffic to VLT Peer B hence minimizing traffic loss.