How does LuCI determine what to list as an "upstream" network in the Overview page?
I have three OpenWrt routers with varying combinations of WAN and Wireguard interfaces through which the internet can be reached. The problem is, it seems to be very arbitrary which interfaces are shown for each router in the "Network -> Upstream" area of the LuCI Overview page. One shows all of them, one shows all of the IPv4 but no IPv6, and one only shows 2 of 3 IPv4 and no IPv6 (despite all of those functioning as expected on all three devices). This isn't the end of the world, because I can get the same information on the Interfaces page, but it's a little weird and unintuitive to me, and it seems to defeat the purpose of a useful "overview".
While I'm on that subject, I suppose I could also mention that static DHCP leases don't show up in the "Active Leases" area, which is strange as well.
It's the output from ubus call network.interface dump, which is the current state derived from various DHCP and Neighbor Discovery chatter between the router and the upstream device.
By "static" do you mean real static assignments on the client device in question, or lease reservations on the router? The former doesn't participate in address assignment via DHCP so there's no "lease" to display; the latter should show up because it's just another lease for which you've explicitly defined the IP rather than letting the router make one up.
How is it decided what is an "upstream" interface from that list (which appears to show all interfaces)? That's the part that doesn't seem to be reliable/complete.
Ah, OK. I just realized that while I had set static leases in OpenWrt for a couple of PCs, they were also set as static within the PC settings as well, so the static leases were redundant and being ignored as you say.
Sorry, I still seem to be missing some aspect of this...
I have Wireguard interfaces that show up in the Upstream area on some of my devices but not on others, despite the tunnel being up and usable in each case, and internet access available through them (based on routing policies, i.e. privacy VPN for most traffic, WAN bypass for some, VPS tunnel for some other).
You mean it must carry a default route (0.0.0.0/0 or ::/0) in the main routing table?
That's frustrating... I wonder if there would be any developer interest in tweaking that behavior to either include interfaces with default routes in other tables, or to add the ability to manually define which interfaces that appear in that section.
One of my Wireguard tunnels connects to a VPS that gives me a stable public IPv4 address and full IPv6 (leased prefix). It's never used as the primary default route on my routers, but typically for a single VLAN and also site-to-site connectivity, so I use a secondary routing table to manage that.
It's annoying when it doesn't show up in that Upstream section because that's the quickest, most concise summary of what's going on with my WAN and my tunnels all in one glance.
