Wireless access point signal dropping constantly

I have made a MoodleBox v4.5.1 by downloading the image and writing to a 16 GB SanDisk microSD card. But it is practically unusable. At the rate of 10-15 seconds the signal drops.

The syslog writes messages like these roughly at the same rate:

[Time stamp] moodlebox hostapt: uap0: STA xx:xx:xx:xx:xx:xx IEEE 802.11: deauthenticated due to local deauth request
dito disassociated

It happens in a Pi 4 B and Pi 3 A+. No other models to test, but a colleague just tried, on a 3 B, run in to the same problem. He initially thought if the card is made with the writer supplied on RPi web base, then it works. But more testing showed that it is not true.

Could it have a connection to the "regulatory country"? I always use the Default Switerland and channel 11.

Edit: Found this related discusion: https://discuss.moodlebox.net/d/287-moodlebox-400-wifi-connections-drops-after-each-request/21.

Nicolas

Thanks for confirming that the "regulatory country" can't be the cause. It sounds illogical for me too. But that is what https://discuss.moodlebox.net/d/287-moodlebox-400-wifi-connections-drops-after-each-request/21.

How about the "hostapt: uap0: STA xx:xx:xx:xx:xx:xx IEEE 802.11: deauthenticated due to local deauth request" messages? There are lot of them alternating with the opposite message "associated" (from memory, not at the MB as I write).

@ralfkrause

Interference with other wireless transmitters is the first thing I investigated. I have an app called WiFiAnalyzer installed on an Android smpart-phone and tried changing the wi-fi channel of the MB. It was a pain though, since the connection was breaking all the time. At the "site" I was trying to get this going, there was no cable Ethernet!

And since then I have tried the same thing in many different places with widely ranging wireless landscape. Even a colleague is trying in a completely different place. I am a noob in wireless transmissions and drivers. My web searches always take me to such topics. Here are some:

• https://forums.raspberrypi.com/viewtopic.php?t=269221

• https://forums.raspberrypi.com//viewtopic.php?f=29&t=269390&p=1635288#p1635288 (linked from the above as the solution)

I must admit that in half of the situations I tried there was another access point in less than 1m distance. But in the other no access point in the vicinity and no wireless noise as monitored by the WiFiAnalyzer. And the MB, in fact there are two, a RPi 4 B and a 3 A+, both never works longer than a minute, then drops the connection. The client reconnects after may be 10 sec. That repeats all the time.

BTW, my colleage and I tried different client devices.

ralfkrause

When I do it, there is no (Windows) GUI. I ditched Windows 20 years ago. Since then my "desktop" (also laptops) is Linux. (Will add that to my profile.) Therefore, it goes like this:

# dd if=moodlebox-4.5.1.img of=/dev/mmcblk0 bs=4M && sync

My colleague has discovered the GUI program (on his Windows) though.

We two work at two different places and on different hardware; I have a 4 B 4G and a 3 A+, he has a couple of 3 B and 3 B+. I don't know what he sets for SSID etc. But the fact is he connects an Android smpart-phone and a PC with a USB wi-fi adapter to the MN. So the SSID/Password pair must be right.

Initially, he even claimed that SDs made with Windows (on the GUI above) work. At a closer look they didn't.

Then he made a new set of SDs and testing one. The test being connect the MB to a router, connect the client to MBs wi-fi and watch whether wireless signal fluctuates and/or it gets disconnected and reconnected, all the time while watching a YouTube video on the client.

Edit: Now comes to my mind, does the WPA standard setting of the client matter? There are all sorts of WPA2, WPA3 with additions like Personal, Extended, ,etc. Or put it another way, what is the correct setting for the WPA? You have no problems with MB v4.5.1, I assume.

Nicolas

Thanks for joining! First of all, I was using that 'dd' command for years. And my colleague was using the Imager on Windows, having the same problem.

Let's forget all that. Installed the RPi Image from the Debian repos and rewrite a SD. (Side note: Obviously it downloaded the image again. Now OK, but not OK when I am at the site.) No big difference. Immediately I felt that the wi-fi connection was brisk. Then started disconnecting again. May be a bit slower, every minute instead of every 30 sec, but it fails, that repeatedly. (The network manager reconnects, works for less than a minute, then down.) The network manager widget tells a lot. The wireless "cone" get smaller, then vanishes and the small wheel starts to turn.

I debug with Conky and bmon. Sie screen-shot. The two figures 'wlan0' and 'Current' (marked red) vanish and reappear.

ralfkrause Changing the wi-fi channel is the first thing I did. Just now even completely switched of the router. (It is a dumb 4G router, I can not switch off the access point, have to pull the power.) Zero access points except MoodleBox as shown by the WiFiAnalyzer. No improvement.

After further experiments through elimination the problem seems to be in my hardware, for some unexplained reason both my RPis, a 4 B and a 3 A+, have a hardware problem.

To make a long story short, my colleague who was testing RPi 3 B and 3 B+ got MoodleBoxes done with stable wi-fi. I borrowed a RPi 3 B and the SD card. Indeed the combination works very stable. With the only caveat that the wi-fi connection speed is only 72 Mbit/s whereas I have seen MoodleBoxes connecting at many times that speed. Other than some sporadically slow ping packets, the speed remains stable.

Now I insert the colleague's SD card in to my Pi 4 B, the old problem, fluctuations in the signal strength and completely dropped connection every minute or so, happens. I insert my SD card to his Pi 3 B, it works as described above.

But that explanation doesn't convince. It means that the 3 A+ I had is also defective. (I don't have it anymore, gave it to a third person.) And my colleague with those "good" 3 Bs, he also ran in to exactly the same problem in his initial tries. He doesn't know what he did differently to make them work.

Any detective work is highly welcome!