B/W Distribution numbers suspect--anyone seeing this?

Report issues relating to bandwith monitoring, bandwidth quotas or QoS in this forum.

Moderator: Moderators

Post Reply
rseiler
Posts: 208
Joined: Sun Dec 15, 2013 12:31 am

B/W Distribution numbers suspect--anyone seeing this?

Post by rseiler »

This isn't happening with my WNDRMACV2, but is on three Archer C7v2s. 1.92 and 1.10.

It's extremely confusing and inconsistent, but I first noticed a problem when the email report BW figures would usually disagree--often dramatically--with the "Sum" line in B/W Distribution. I don't know when the problem started, but it dates back at least a month. And it's often impossible to tell which figure (router or email) is true, as I'll try to explain.

Sometimes the router's numbers are higher, sometimes lower, relative to the emails (I don't mean slightly higher or lower, which would be easy enough to ignore). Why not always lower or always higher--if they're going to differ in the first place? No idea.

But surely related to this is that the router on some days show a very low total and zero usage for devices that I know were there. Not all the devices are zero on those days; statically-assigned ones are not. Most times this is NOT the case--most days, there's a mixture of devices showing usage, but even on those days, the router and email will likely not agree. And sometimes the router shows (probable) normal usage and the email has an extremely low figure for, say, DL. It's just not consistent.

Maybe this is wrapped up with DHCP somehow (router is enabled for DHCP). And I should mention that one thing the Archers have in common is that they all use guest WLAN, but that wouldn't explain the zeros just mentioned above, which are PCs not connecting to the guest WLAN.

Of late, I've been scheduling the routers to reboot at 3AM instead of weekly. That hasn't helped.

ispyisail
Moderator
Posts: 5185
Joined: Mon Apr 06, 2009 3:15 am
Location: New Zealand

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by ispyisail »

Of late, I've been scheduling the routers to reboot at 3AM instead of weekly. That hasn't helped.
In another post somebody suggested that rebooting was causing the problem?

rseiler
Posts: 208
Joined: Sun Dec 15, 2013 12:31 am

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by rseiler »

ispyisail wrote:In another post somebody suggested that rebooting was causing the problem?
For what duration afterwards? It was happening when rebooting weekly, too. And this thread, which seems like it might be related, suggests that rebooting can help:
viewtopic.php?f=12&t=11300

That thread also mentions the Archer, which may be a coincidence, we'll have to see. That it doesn't happen at all on the WNDRMACv2 is interesting but not in itself that helpful, since it's not configured in exactly the same way as the Archers.

rseiler
Posts: 208
Joined: Sun Dec 15, 2013 12:31 am

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by rseiler »

I think I finally cracked the problem. Short version: BW data is being corrupted retroactively. It's not an email problem (they're just following the router), but the emails helped me discover the problem.

Please, someone check me on this.

Here's what I think I know now:

First, it is happening with both models. This was so tricky (you'll soon see why) that I initially missed that.

The router and the emails are showing correct BW (or what look to be correct) if it's the latest day that you're examining. So, for example, if today is Nov 11, the Nov 11 line in the email (I have the email configured for 5 days) and the Nov 11 data in the router will agree.

But when Nov 12 is over and its data is complete, the Nov 11 data now shown in the router is magically different and thus wrong along with the Nov 11 data in the Nov 12 email (which makes sense, since the email is just pulling data from the router).

You can see how easy it was to be thrown off by an email that includes the last 5 days when only the latest day is solid.

I have no idea where (and certainly not why) the router is getting the numbers from for past days, but sometimes they're incredibly wrong (say, 12MB for DL) and sometimes they're more realistically wrong (the type that you wouldn't necessarily notice right away).

rseiler
Posts: 208
Joined: Sun Dec 15, 2013 12:31 am

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by rseiler »

ispyisail wrote:In another post somebody suggested that rebooting was causing the problem?
I can't find the thread. Do you still know where it is?

In light of the new information, I'll experiment more with the possibility.

I wonder if manual reboot and scheduled reboot both use reboot.sh?

schneidler
Posts: 3
Joined: Fri Oct 14, 2016 1:55 pm

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by schneidler »

I wasn't sure if this should be a new thread or a reply to this thread. I don't post very often. But I am seeing this issue as well on my gargoyle router. I have a user on our shared home network who was out of state with all of their devices, but the bandwidth distribution table shows like 500, 700, 400 mb of downloads per day for their iphone for some of the days they were gone (other days show zero, as expected). I am using static IPs with MAC address control enabled, and I confirmed the MAC address of the iphone in question and that it was in the user's pocket in Idaho at the time of the usage in question.

??!!? This is a problem because we all share 40 gigs per month of internet and I have to bill everyone their proportional share based on their usage. But if the usage reports can't be trusted, I don't know how to do that. Please help if you have any insight. Thanks!
-Peter in Alaska

rseiler
Posts: 208
Joined: Sun Dec 15, 2013 12:31 am

Re: B/W Distribution numbers suspect--anyone seeing this?

Post by rseiler »

That does sound like a different issue--unless you're seeing that the most recent day has the right data but then seeing that day's data changing after some amount of time passes and it's no longer the current day (this amount of time varies in a way that I've never understood).

I've only been looking at the bottom-line totals--but it stands to reason that at least some of the devices would be wrong too, once things go wrong.

So the key here in determining if we're talking about the same thing is all in the timing of when you're looking at the data. If you're seeing that impossible usage on the day, I think it's a different problem. After that day passes though, all bets are off, and the history is wildly unpredictable.

Post Reply