Configuring Keep-Alive Timeout for Virtual Link Extension
As part of the vLE link state tracking logic each node hosting a vLE endpoint sends fabric fast keepalive messages every second to its peer node that hosts the other endpoint.
A node considers its peer node down if a keepalive is not received from the remote node within 3 seconds (which is the default vLE tracking timeout). If that happens, the local port is brought down and the vLE state reflects both ports’ down state.
However, in some deployments (for example with frequent link flaps), a three second timeout may not be enough to avoid false positives. Therefore, it can be modified with the following command:
CLI (network-admin@switch) > system-settings-modify vle-tracking-timeout seconds
You can configure a value from 3 to 30 seconds. (Note that the fast keepalive transmission frequency remains one second. Only the timeout value is adjusted to the configured value.) The configured value can be checked with the system-settings-show command.
Note: The default 3 second keepalive timeout is automatically configured when vLE link state tracking is enabled. When it is disabled, the standard fabric message timeout interval is 21 seconds as shown for the three leaf switches below that don’t use tracking:
CLI (network-admin@switch) > fabric-node-show format name,keepalive-timeout
name keepalive-timeout
-------------- -----------------
leaf-1 21
leaf-2 21
leaf-3 21
switch 3
CLI (network-admin@switch) > system-settings-show format vle-tracking-timeout
vle-tracking-timeout: 3
When tracking gets disabled, the fabric keepalive timeout goes back to 21 seconds:
CLI (network-admin@switch) > vle-modify name vLE no-tracking
CLI (network-admin@switch) > fabric-node-show format name,keepalive-timeout
name keepalive-timeout
-------------- -----------------
leaf-1 21
leaf-2 21
leaf-3 21
switch 21
CLI (network-admin@switch) > system-settings-show format vle-tracking-timeout,
vle-tracking-timeout: 3
When it is necessary to change the administrative vLE keepalive timeout value, the fabric keepalive timeout reflects the new value when tracking is enabled, as shown below:
CLI (network-admin@switch) > system-settings-modify vle-tracking-timeout 30
CLI (network-admin@switch) > fabric-node-show format name,keepalive-timeout
name keepalive-timeout
-------------- -----------------
leaf-1 21
leaf-2 21
leaf-3 21
switch 30