Help understanding Tx and RX not updating

Hey, after managing to set up my VLANs, I've been playing around with gathering stats from the interfaces on OpenWrt (using luci-app-statistics to export as prometheus data). I currently have 2 VLANs on my LAN (15 and 25) and another one on my WAN interface (12 as my ISP forces me to do so).

To try and make sure I was interpreting the TX and RX values correctly, so on the wan's perspective, tx == upload, rx == download, I downloaded several 1GB files expecting to see the stats rising, which did not happen. Any idea why?

Router: TP-Link Archer C6 v3
OpenWrt version: OpenWrt 23.05.0 r23497-6637af95aa / LuCI openwrt-23.05 branch git-23.236.53405-fc638c8

Edit 2:
The download was performed on a laptop directly connected to the router, on VLAN 25 (untagged port).

If you enabled (software- or hardware) flow-offloading, the kernel will never get to see large chunks of the traffic, as it bypasses netfilter.

1 Like

Thanks! That's indeed the case here. So the trade-off is better performance and no cool metrics vs worse performance with cool metrics? I guess I'll just let it consume more power.

Before if I start increasing my CO2 footprint, is there an alternative that allows me to keep the better performance with accurate figures for Rx/TX values?

Not really, either the kernel keeps taps on everything or you let parts of the traffic bypass it. The carbon footprint, at least as long as you stick to this hardware (and don't need faster/ more power-hungry devices), probably won't change in a measurable way, contrary to more modern ARM based SOCs, mips based ones used in routers don't really employ frequency scaling or lower power modes (and even on ARM based ones, the impact is rather small).

Thanks for clarifying. I thought enabling hardware offloading drastically decreased CPU usage, but to be honest, I haven't measured anything. I'll untick the box.

It does reduce CPU usage, a lot, but that doesn't result in less electricity usage, nor less heat. These mips SOC use rather little electricity (and they aren't fast by today's standards), but their usage is rather steady, regardless of the CPU utilization/ load. They are descendants of mips workstations of the early to mit 90s (SGI and friends), they never got new features beyond that (aside from process shrinkage, bumping the frequency and adding modern I/O IP blocks), power reduction has not been in their schedule.

Got you. Will unticking this box affect overall performance? I sometimes have days in which 3/4 people are working remotely plus 5/6 phones and TVs hogging YouTube.

Depends on your wan speed and other usage, test it.


This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.