Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: The Natural Philosopher <tnp@invalid.invalid>
Newsgroups: comp.sys.raspberry-pi
Subject: Re: Spontaneous locale change on Bookworm
Date: Sat, 28 Sep 2024 11:55:12 +0100
Organization: A little, after lunch
Lines: 92
Message-ID: <vd8nai$17d93$1@dont-email.me>
References: <vcmq6m$1l88o$1@dont-email.me> <vd8i2e$16f67$1@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Sat, 28 Sep 2024 12:55:14 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="92a90caf65006c84dcecd3f39c660d74";
	logging-data="1291555"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX19ZbZ4AobiftkbgdbPqcrS/gaSOwF8KQ/k="
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:/l0nD2Dk0SyX9HsnsXUg7xbnIjk=
In-Reply-To: <vd8i2e$16f67$1@dont-email.me>
Content-Language: en-GB
Bytes: 4871

On 28/09/2024 10:25, Deloptes wrote:
> bp@www.zefox.net wrote:
> 
>> This morning, after a few minutes' use of my Pi5 running Bookworm,
>> the WiFi connection abruptly dropped. Couldn't bring it back up,
>> so I rebooted the access point. No luck. Then I rebooted the Pi5
>> and noticed something about rfkill stopping wifi in the boot messages.
>> At this point the last reboot was several days past, wifi hadn't been
>> touched or given any trouble.
> 
> do you have NTP configured and do you have networkmanager in use?
> 
> If not get them running and see if it solved the issues.
> 
> (been there seen it all)
> 
> BR

Hard to get NTP to run without a network connection

My PICO pi Ws dont even have calendar clocks, and they connect OK.
Let me list all the things that people have said fucked up their wifi on 
PI Zeros Ws...and other Pis...

- power saving mode (one existing Pi Zero  W had it, the other did not. 
No difference connecting to same wifi access point)
- different wifi chipset (eventually I discovered all my pis were 
running the same chipset)
- interference from HDMI (Headless. HDMI disabled)
- random allocation of MAC addresses.  (disabled - no difference)
- No NTP configured (you) [always configured by me anyway]
- network manager not running (you) [it's part of the default bookworm 
installation]
- various odd 802.11 modes on their router.. (my other two Pis connected 
to my main wifi point just fine. I changed the Pi to run on a different 
one and the problem persisted)
- crappy power supply (me, and others) [ changing the power supply has 
resulted in no drops since connected to any WiFi access point]

(pifi2:~ $ uptime
  11:52:12 up 1 day, 22:49,  2 users,  load average: 0.00, 0.00, 0.00

That now running against  the access point that it had when dropping out 
at least once every 24 hours

I will leave it a few more days to make sure...)

As you can see I went through *every single one*.

And fixed it one way or another, (I didn't worry about ntp: If I am 
running linux on a network there is no way that wont be enabled by me, 
and Bookworm defaults to network manager anyway, so  that  is also  default)

The *only* one that actually fixed the problem and made Bookworm Pi Z 2W 
as rock stable as all my other Pi's was putting a larger PSU on it.

I am not saying that other people's problems are the same cause, but I 
am saying that hairy ass guesses are simply no substitute for trying 
*all* of the above systematically.

Because the  log files - systemd or otherwise -, are simply full of 
useless detail. There to help the programmer debug his code, not to help 
Joe Ordinary find out why his wifi is fucked.

For us, its a matter of trial and error until a stable system is arrived at.

Now my *conjecture* is that both a Zero 2W,  a Pi 5 and indeed Bookworm 
on it, use more power. And a marginal PSU will fail more readily, and 
the wifi is the first point of failure

Because this problem is not in general related to Bookworm on any other 
platform. And the wifi chips themselves are pretty much the same

But in the end, my original thread was not 'give me your hairy assed 
guesses' but 'how do I solve this issue algorithmically'

And the answer was that the log files were useless, and trial and error 
and eliminating one issue at a time was the only pragmatic approach.

And everyone had a 'this worked for me' fix and none of them were the same.

Now I too have a 'this worked for me' fix, but I *also* have a pragmatic 
approach to bug hunting this one.


-- 
"Anyone who believes that the laws of physics are mere social 
conventions is invited to try transgressing those conventions from the 
windows of my apartment. (I live on the twenty-first floor.) "

Alan Sokal