Traffic spike fix from 384.14_2 was reverted in 386.1

[RMerlin]

In all those years, I have never been able to reproduce the issue here.

I have one or two things I could try, but I'd need someone who is able to easily reproduce the issue to test a patched rstats.

 

[DeepAnger]

I volunteer. Last 24 Hours always displays impossible values. And here are 2 screenshots taken a few seconds apart.

 

[Jeffrey Young]

It may be useful for Merlin to post current wan setup, including any vlan tagging for those having issues. I have seen where flow cache can play havoc with this. Although in that case, it was a case of not seeing any data. Turning off fc solved the issue there.

 

[DeepAnger]

More information.

 

[Wistuplu]

Hello,

DeepAnger, thank you for sharing input.
May I suggest that, in future reports, one tries to actually show the spike. I mean, in the screencapture above, the graph for "Internet Connection (WAN)" is flat. In my experience, one of the other graphs (e.g. "Wireless 2.4" or "Wireless 5") actually shows the ghost spike. In my case for AC86U, it has been mostly one of the "Wireless" graphs. It would be interesting to know about various experiences: Is the spike visible in "Wired", "Wireless 2.4" or "Wireless 5" ?

Below, an example from latest ASUS beta, 9.0.0.4.386_41994. Note how the graph there has a green title (e.g. "Wireless (2/4GHz)") which reflects the choice of WAN / Wired / 2.4GHz / 5.0 GHz. So that closed-source component seems to have changed. Also, when I reported the issue to Asus, they directed me to the 41994 Beta. I don't know if it's a strategy of just trying the latest available, or if 41994 was really supposed to bring improvements in this area.

Anyway 41994 is still prone to spikes. The spike comes usually after between 10 to 30 seconds in my case. This time it came after a reported approx. 20MB of real traffic, based on the total displayed in the table around the time the first spike appears, before figures got all crazy.

I will most probably be back to the Merlin firmware soon. I thought it was interesting to illustrate the behaviour again with the official firmware while I still could.

Thank you
Cheers

 

[DeepAnger]

I did before there:
https://www.snbforums.com/threads/asuswrt-merlin-386-1-is-now-available.69783/page-16#post-657994
But I usually don't care about Wireless because more than 99.99 % of my traffic comes from one or 2 PCs in Gigabit Ethernet.
But even when I select Wireless 2.4 or 5 graphs are just the same.
My long VDSL2 line gives me 23 Mbps down and 3 Mbps up at best. So anything above 3 MB/s is impossible (except for WiFi). And whatever your connexion type, 614891468983371.50 KB/s is impossible too.

 

[DeepAnger]

Wow, I never checked Wired before but (in my case) that's where the problem makes the graph crazy.

I have a VDSL2 "livebox" connected to the RT-AC68U (Gigabit Ethernet) connected to a CISCO switch (Gigabit Ethernet) where are connected the rest of my LAN devices.

 

[Morac]

I get at least one spike a day, sometimes more, so for me it’s fairly reproducible.

In my case the only thing that shows the spike is the WAN link. There’s nothing on the wired or wireless graphs. Based on the spike number it’s fairly obvious that 2^64 - 1 values are being written. I‘m guessing there are negative values being written to an unsigned long value, though I have no idea why that would be.

 

[ech]

I knew it would happen when I upgraded to 386.1, and indeed, within less than 3 hours, I see a spike just like Morac's.

Happy to test a patched rstats.

 

[RMerlin]

I need router models. Code has to be compiled for specific routers.

 

[FTC]

I get spikes in my RT-AX88u WAN graph only, almost every time when WAN is disconnected (manually through GUI or by the ISP for maintenance for instance). Graph is similar to Morac's.

 

[Morac]

I’m using a RT-AC88U.

 

[ech]

I'm on an RT-AC68U

 

[Wistuplu]

If you are interested, I can test on AC86U. Note that I'm in a different timezone, CET in Europe. So, it might slow down interactions.
I usually get the spikes within 10 to 30 seconds. So, they are super easy to reproduce.
Note that tonight, I had almost zero traffic on wireless, and the spike didn't come as fast as usual, despite significant traffic on a wired client. I feel my graph usually get crazy due to wireless traffic (like as soon as 20MB traffic on Wireless). Other persons above have different experiences, on different models.

Thank you very much. Regards

 

[dev_null]

For anyone interested, I posted my Vnstat on Merlin (with UI) process here: https://www.snbforums.com/threads/vnstat-on-merlin-cli-ui-and-email-data-use-monitoring.70091/

 

[Morac]

I got a response from ASUS about my bug report that they are escalating to engineering. I guess I’ll see what they say.

 

[Morac]

I got a response from ASUS about my bug report that they are escalating to engineering. I guess I’ll see what they say.

The response was to try installing 9.0.0.4.386.41994 beta, which @Wistuplu said also has the problem. They said if the problem still exists to use the feedback mechanism in the GUI.

So as I expected my bug report didn’t make it to the right people.

 

[Wistuplu]

as I expected my bug report didn’t make it to the right people

Just not yet ;-) .
RMerlin mentioned he had never seen the issue. So, even though it is obvious for those who experience it, like you or me, it is apparently not universal.
I'm not surprised the helpdesk wants the customers to reproduce the issue with the latest release before passing it to the real developers.

 

[Morac]

Just not yet ;-) .
RMerlin mentioned he had never seen the issue. So, even though it is obvious for those who experience it, like you or me, it is apparently not universal.
I'm not surprised the helpdesk wants the customers to reproduce the issue with the latest release before passing it to the real developers.

They didn't say reproduce it and get back to them in the same case. They said if I can reproduce it, use the feedback page in the UI.

 

[Wistuplu]

2cts...
My interest in Traffic Analyzer / Traffic Monitor / Real-time is related to the display of the "instant" traffic / throughput.

I only just realize that the Android app also provides the information, albeit in a less convenient/clear format. And it seems to not be impacted by ghost traffic spikes and inadequate scaling. I share it here in the hope that this might help others.
Take care

BTW: When using the feedback page in the UI, one can reference a case number. I don't know if this is actually working, and/or if it allows reactivating a "sleeping" case.