TL;DR: On NVIDIA/Mellanox switches running Cumulus Linux, list existing VLANs with nv config show, add a new VLAN with nv set bridge domain br_default vlan <id>, then apply and save the configuration so it survives a reboot.
Recommended action:
Connect to the switch console (or SSH in as the
cumulususer) and review the current configuration:nv config show
Look under
bridge → domain → br_default → vlanfor the list of VLAN IDs already trunked on the bridge.Add the new VLAN to the default bridge domain. Replace
XXXwith the VLAN ID you want to add (for example,60):nv set bridge domain br_default vlan XXX
Apply the pending change to the running configuration:
nv config apply
Persist the change so it survives a reboot:
nv config save
Repeat the same sequence on the second switch in the pair so both peers carry the new VLAN identically.
Why:
NVIDIA Cumulus Linux uses NVUE (nv commands) to manage configuration. Adding a VLAN to br_default makes it available across all bridge member ports — i.e., the trunks between switches and the server-facing links that are members of the default bridge. nv config apply activates the change in the running config, but only nv config save writes it to startup configuration; without the save step, the VLAN disappears on the next reboot.
Going forward:
Always run nv config show first to see the existing VLAN list before adding — this avoids reusing an ID that's already trunked. Apply and save on both switches in a redundant pair within the same maintenance window so they stay in sync. After the change, validate from a host on the new VLAN that it can reach its expected gateway/peers.
Optional details:
If a port should carry only specific VLANs rather than every VLAN on br_default, you can scope membership at the interface level (nv set interface swpX bridge domain br_default ...). For most server-facing access/trunk scenarios on a DataON-shipped pair, adding the VLAN to the bridge domain is sufficient because the server links are already members of br_default.