Luci 19 - graph noise when wireless disabled?

Status / realtime graphs / wireless

Anybody else have permanent graph noise when wireless is disabled?

This is absence of signal... not noise.

Under the graph it says Blue for signal and pink for Noise so shouldn't they both be Zero with a empty graph as wireless is disabled?

Maybe it’s my eyes, but as I see it on my phone, -255 dBm is pretty close to zero (roughly a billionth of a billionth of a billionth of a Watt).

Seems to be a graph-scaling problem, not a data problem.


But does everyone have this when wireless is disabled or just me?

yes i have this, 18.06.4 archer c5-v2 ( wireless disabled )

EDIT: nope wireless was enabled...


Thanks for confirming that , i have a Linksys EA6350

@bill888 has it with 18 and 19 on a BT HH5A router but not on LEDE 17.

That's 3 confirmations on different routers so i'm guessing we all have it and can be ignored , just seems odd to have it wasting resources when it's disabled and not being used.

Showing other WiFi signals in your Area would be useful as noise. :grinning:

I have not seen any wifi driver report a varying "noise" number that would suggest that RF noise is actually being measured. Usually it is a constant regardless of RF conditions.

1 Like

But as Wireless is disabled shouldn't it be inactive along with everything else?

-255 is an "interesting" number, if you think about it.

On a logarithmic (log) scale such as dB, there is no "zero", only smaller and smaller. -255 is as close to "no power" on a log scale that an 8-bit, signed number can represent


But as wireless is disabled shouldn't all logging be disabled as Bill doesn't have it with 17?

v17 is likely significantly different as far as LuCI and its plug-ins go. Purely a guess, but something like this probably isn't very high on the to-fix list.

There are some usability concerns around "hiding" the graphs if the interface is disabled, such as if you've recently disabled it, you can't see the past history (which might not have been non-zero).

If I were to fix it (no, that isn't an offer), I'd probably address it by fixing at least the low end of the power axis. I'm not a big fan of self-scaling axes when the data range varies relatively quickly with time, You never know without looking carefully at the axis if that visually "big change" is 3 dB or 30 dB. Or, as the graph tricked you (and me, until I looked closely), if that "big level" is big or not.

But then why not carry on logging everything else to see past history , I must have different reasoning of disabled? :grinning:

Logging the statistics associated with an interface and an interface being "up" are two, disconnected processes.

Knowing that something is "zero" for a period of time is, many times, valuable information. Turn logging off, can you tell from the data if logging failed, or if the interface was down? Think about running-average calculations -- you'd have to special-case the "no data" sections, rather than knowing you're getting data every second (or 10 seconds, or ...). Logging is relatively "cheap" as far as resources go for most things1.

1 Properly done and at a reasonable level of detail. A recent, notable exception wat the detailed statistics being logged from, as I recall, the QCA switch chips, that caused excessive load on a bus designed for occasional control signals, not streaming of diagnostic data. See further, commit 107dc4326c

When you turn off your cable/satellite TV box your screen will briefly inform you that there is " No Signal " you don't expect it to be permanently on flashing away telling you something you already know? :grinning:

Different users, different worlds for those two interfaces. Your TV is still collecting the data and it is likely available on a diagnostic screen.

Easy solution -- don't click on the tab.

More challenging solution -- learn enough LUA to change the graph axes.

Yeah, this is not so much a "scale" issue or resource issue as I see it....

Essentially, as wireless is disabled... if anything ( the average user ) should just want it to be blank.... or say disabled. Which is a minor "nice to have"....

But yes, I can see also how the average/intermediate user might be a little confused by the graph output....

Q.) Something has changed since 17 is it a new feature or a bug?

A.) You can answer that yourself by not looking at it.

:roll_eyes: :grinning:

Or the advanced user.

Were the scale to at least be fixed (static, constant) at the bottom to, say, -100 dB, or even clamped to never go below that, the -255 input value would result in an empty graph.

1 Like

The amount of discussion spent on this issue probably by far outweighs the effort required to implement a fix, so - patches welcome.