Sailfish on Pro1-X is possibly causing my wifi AP to drop packets?

[ducksoup 110]

I am in the early stages of tracking something very strange indeed.  I've been getting alerts for a few days from my own internal network monitoring (Nagios) that my primary wifi AP is often dropping packets on its ethernet interface.  Since I've been busy with other things (not the least of which is the Pro1-X itself), and the packet loss isn't enough to cause problems, I didn't look into the situation carefully until last night.

Once I pulled up my reachability graphs, I realised that the issues seem to have started at exactly the same time I received the Pro1-X:

There are a couple of immediately very unusual things about this correlation.  First of all, the Pro1-X was connected to my guest wifi for the first day, which is a physically distinct AP (and showed no problems), and second, the observed packet loss is on the ethernet side of my primary AP, not the wireless side (which is not to say there isn't any loss on the wireless side; I'm not testing this side, and perhaps should be).

But I have tested repeatedly, and found that any time I turn the Pro1-X off, the problem disappears:

(powered down at 15:40)

(booted into EDL mode at 14:15)

Further, I noticed that there were a couple of long periods when the problem seemed to go away, but times that I knew I was using the Pro1-X, on wifi.  Further testing has revealed that I'm seeing this problem only when the Pro1-X is either charging or discharging its battery, and not when it is sitting on the charger with a fully charged battery.  Also, this is happening only with the OS booted, not when the device is in fastboot or EDL mode (where I imagine the wifi radio is powered off?).

Now, theoretically no matter what the Pro1-X is doing on the wifi end of things, my AP shouldn't be dropping packets on the ethernet end of things, but it's not a particularly good AP, so this has got me wondering if the Pro1-X is doing something with its wifi radio that is upsetting my AP enough that it gets too busy to respond to every ICMP request on the ethernet side.  And, for whatever reason, this something seems to be happening only when the battery is charging or discharging, not when it's steady.

At the moment I'm making a backup of my Sailfish setup, and I plan to install Ubuntu Touch and perform some further tests and see if the problem follows.  I expect that it will, because I do have packet loss in my reachability history going back to when I first received the device, and I had it running the factory Android build for most of the first day.  I do note, however, that it looks like the problem has been worse more recently, so Sailfish may be a factor.

There seems to be nothing unusual on the ethernet layer or higher coming off the Pro1-X, but I do not have the equipment or expertise to investigate this further down into the 802.11 and radio layers.  But perhaps someone else does?

I'll add more findings once I get some experimentation done with Ubuntu Touch.

 

Edited August 14, 2022 by ducksoup

[ducksoup 110]

Okay, it looks like this problem actually does not appear after I flash Ubuntu Touch.  (I do, however, now have the problem that the screen goes black on UT whenever I close the keyboard, but that's unrelated!)

Not sure at this point why I saw such variability on my AP going back to the day I got the Pro1-X.  But it does appear that at least the worst examples of the problem are exclusive to Sailfish.  I have updated the topic on this post appropriately, and will continue to investigate!

 

[ducksoup 110]

After a couple of days of playing around with mostly Ubuntu Touch, but also a little bit of stock Android. I've just flashed Sailfish OS back onto the Pro1-X, and sure enough, immediately my AP is dropping packets on the ethernet side again.  Will do further investigation.

[matf-kabouik 414]

That's interesting. I'm experiencing issues with my home network on other devices as well since a few days. I don't remember any of it in the first days i've used Sailfish, though, but maybe I just didn't pay attention. My laptops are losing connection to the WiFi network regularly, but surprisingly the wired devices don't seem to be affected.

I don't know if it is related, but it definitely could: ssh over WiFi to the Pro1x running SailfishOS is *very* unstable, and gets more unstable with time, until it eventually just disconnects. Pretty sure this could be the same root cause.

Edited August 16, 2022 by matf-kabouik

[ducksoup 110]

I talked to a friend with a lot more experience in these subjects than I, and he suggested using something like a HackRF One to look at the radio transmissions themselves to see what's going on.  That's a little out of my pay grade (and I don't own a HackRF One), but perhaps someone else has those skills.

Unfortunately this isn't a local friend, so I can't easily bring him over to have a look himself.

[liquidvx 8]

20 hours ago, matf-kabouik said:

I'm experiencing issues with my home network on other devices as well since a few days

Ditto. After a couple days my Pro1X with SailfishOS seemed to cause wifi problems and yesterday I had to reboot the router twice. Turned off the device last night and haven't had network problems since.

The issue I'm experiencing reminded me of the Google home/Chromecast bug that was killing wifi by sending bursts (< 100k) of MDNS multicast discovery packets when the device exits sleep

Quote

TP-Link says the longer the device sleeps, the "larger this packet burst will be."

 

The linked article also mentions "phone" so maybe mobile devices could contain the firmware bug but hard to believe since it was like 5 years ago. If it is the same bug then it would explain why the network problem appears to after some time. I will power on my Pro1X and see if I can get more info from router diagnostics.