From Newsgroup: comp.sys.raspberry-pi

After a month or so of working well, the wifi has again started acting
strange after a series of software updates on a Pi5 running Bookworm.

Now the most obvious problem shown in journalctl seems to be authentication. Entries reporting "no secrets" and authentication failed seem to be the norm.

At the same time, sometimes wifi will work long enough to let half or so
of a homepage load.

One report on the web suggested that certain characters in passwords
were not handled correctly in bookworm. The only "special" character
I use is a period (.), which according to all reports I can find is
legal for ssid passwords.

Those clues don't square with each other very well, but they're all I have
for now. The AP works fine with other hosts and is the strongest signal
by about 20 dB. A second host in the same location as the Bookworm machine
has no trouble connecting.

Thanks for reading, and any ideas.

bob prohaska

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 16/06/2025 at 23:26, bp@www.zefox.net wrote:

After a month or so of working well, the wifi has again started acting strange after a series of software updates on a Pi5 running Bookworm.

Now the most obvious problem shown in journalctl seems to be authentication. Entries reporting "no secrets" and authentication failed seem to be the norm.

At the same time, sometimes wifi will work long enough to let half or so
of a homepage load.

One report on the web suggested that certain characters in passwords
were not handled correctly in bookworm. The only "special" character
I use is a period (.), which according to all reports I can find is
legal for ssid passwords.

Those clues don't square with each other very well, but they're all I have for now. The AP works fine with other hosts and is the strongest signal
by about 20 dB. A second host in the same location as the Bookworm machine has no trouble connecting.

Thanks for reading, and any ideas.

bob prohaska

Same problem with a Pi Zero 2W Rev 1 Bookworm.
I check the link every minute and restart NetworkManager if link not there.
It happens once or twice a day, but sometimes up to 4 times.
No regularity to it, though.
No problems with Pi3B+ on same circuit.
--
Chris Elvidge, England
I WILL NOT MOCK MRS. DUMBFACE

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On Mon, 16 Jun 2025 22:26:46 -0000 (UTC), bp wrote:

After a month or so of working well, the wifi has again started acting strange after a series of software updates on a Pi5 running Bookworm.

I have a wi-fi card in an older (Core i7) machine, running hostapd. That
works fine, unless I transfer multi-gigabytes of data to my laptop. That
gets the wi-fi into a stalled state. At that point I remember I shouldn’t have done that, and switch to an Ethernet connection to continue the
transfer.

When I try again some hours later, the wi-fi seems to have cleared its trouble.

Not sure if this is related, but just thought I’d mention it; are you transferring lots of data?
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

Lawrence D'Oliveiro <ldo@nz.invalid> wrote:

On Mon, 16 Jun 2025 22:26:46 -0000 (UTC), bp wrote:

After a month or so of working well, the wifi has again started acting
strange after a series of software updates on a Pi5 running Bookworm.

I have a wi-fi card in an older (Core i7) machine, running hostapd. That works fine, unless I transfer multi-gigabytes of data to my laptop. That gets the wi-fi into a stalled state. At that point I remember I shouldn’t have done that, and switch to an Ethernet connection to continue the transfer.

When I try again some hours later, the wi-fi seems to have cleared its trouble.

Not sure if this is related, but just thought I’d mention it; are you transferring lots of data?

Generally, the problem seemed load independent.

I say "seemed" because shortly after posting my gripe I went back to
the Pi and tried to run ifconfig wlan0 up, whereupon the response was "...prohibited by rfkill..". That I recognized as a localization problem.
I hadn't changed localization, it must have been an update.

I'd previously suspected as much and gone through the hoops of setting
country, keyboard, wifi, etc. more than once to no avail, but I did it
again and this time the wifi started working.

In the meantime I've moved the Pi5 to a location that allows wired
networking. It's much less convenient, but for now I'll keep it there.

Thanks for writing!

bob prohaska

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 16/06/2025 23:26, bp@www.zefox.net wrote:

After a month or so of working well, the wifi has again started acting strange after a series of software updates on a Pi5 running Bookworm.

Now the most obvious problem shown in journalctl seems to be authentication. Entries reporting "no secrets" and authentication failed seem to be the norm.

At the same time, sometimes wifi will work long enough to let half or so
of a homepage load.

One report on the web suggested that certain characters in passwords
were not handled correctly in bookworm. The only "special" character
I use is a period (.), which according to all reports I can find is
legal for ssid passwords.

I have a period in my SSID pasword and all my Pis work fine

Those clues don't square with each other very well, but they're all I have for now. The AP works fine with other hosts and is the strongest signal
by about 20 dB. A second host in the same location as the Bookworm machine has no trouble connecting.

Thanks for reading, and any ideas.

bob prohaska

I had a Pi zero problem that sounds suspiciously like this. A better PSU
made it disappear.

I have a sneaking suspicion that a lot of Pis loaded up with whatever
are now actually on the limit with respect to PSU current. Wifi takes a
lot of power when actually transmitting data.

Friend had a Pi5 that wouldn't connect to his disk, via USB but would
connect to mine.

Powered hub sorted it...
--
“The fundamental cause of the trouble in the modern world today is that
the stupid are cocksure while the intelligent are full of doubt."

- Bertrand Russell


--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 17.06.25 11:08, The Natural Philosopher wrote:

On 16/06/2025 23:26, bp@www.zefox.net wrote:

After a month or so of working well, the wifi has again started acting
strange after a series of software updates on a Pi5 running Bookworm.

Now the most obvious problem shown in journalctl seems to be
authentication.
Entries reporting "no secrets" and authentication failed seem to be
the norm.

At the same time, sometimes wifi will work long enough to let half or so
of a homepage load.

One report on the web suggested that certain characters in passwords
were not handled correctly in bookworm. The only "special" character
I use is a period (.), which according to all reports I can find is
legal for ssid passwords.

I have a period in my SSID pasword and all my Pis work fine

When building a Phoniebox for one of our grandchildren, I learned about
the "wpa_passphrase" command. It generates a 256-bit PSK from the SSID
and the ASCII passphrase and dumps it as a hex string, eg

$ wpa_passphrase abc defghijkl
network={
ssid="abc"
#psk="defghijkl"

psk=6db784a0ea8130c5ff07e690923f9af50cfea5c0776b607bea9bc4478c41f8dd
}

This should
a) get rid of any special character issues
b) speed up the connection establishment.

I successfully tried it on my kubuntu laptop as well.

Josef
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

The Natural Philosopher <tnp@invalid.invalid> wrote:

On 16/06/2025 23:26, bp@www.zefox.net wrote:

One report on the web suggested that certain characters in passwords
were not handled correctly in bookworm. The only "special" character
I use is a period (.), which according to all reports I can find is
legal for ssid passwords.

I have a period in my SSID pasword and all my Pis work fine

I had a Pi zero problem that sounds suspiciously like this. A better PSU made it disappear.

I have a sneaking suspicion that a lot of Pis loaded up with whatever
are now actually on the limit with respect to PSU current. Wifi takes a
lot of power when actually transmitting data.

Friend had a Pi5 that wouldn't connect to his disk, via USB but would connect to mine.

Powered hub sorted it...

In the course of exploring the wifi problems I moved the Pi5 to a new
location where it could use a wired ethernet connection. On reboot I
saw a message saying that "USB booting requires a 5V 5A power supply"
or words to that effect. I think the message is new, it's unclear if
it's a warning or merely informational. At the same time I added a
2nd monitor, which led to some confusion (mine) during startup so it
took a couple of tries, ending with both monitors working correctly.
Whether the failures were boot problems or just me looking at the wrong
screen is not apparent.

Right now there's a voltmeter connected to the ground and +5V pins on
the GPIO header, Wifi is on, the meter reads 5.08V. Turning wifi off
seems to make no difference. Turning wifi back on caused a momentary
dip to 5.06V which recovered within one second.

If that's a PSU problem it's mighty subtle.

Thanks for writing!

bob prohaska

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 17/06/2025 16:08, bp@www.zefox.net wrote:

Turning wifi back on caused a momentary
dip to 5.06V which recovered within one second.

If that's a PSU problem it's mighty subtle.

It should not have dipped *at all*

Have you got a storage scope you can put on it?

My guess - without massive confidence - is that there is a momentary
brownout of a lot less than 5.06V.
--
Labour - a bunch of rich people convincing poor people to vote for rich
people by telling poor people that "other" rich people are the reason
they are poor.

Peter Thompson

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 16/06/2025 23:26, bp@www.zefox.net wrote:

After a month or so of working well, the wifi has again started acting strange after a series of software updates on a Pi5 running Bookworm.

I've had far more problems with Bookworm since it moved to Network
Manager than with Bullseye and dhcpcd. I had to wind back a unpdate to
one and only one of my Pi 5s at the beginning of the year, due to unreliability, but whatever it was went away at the next update. Luckily
I have two root partitions on the nvme, so I can upgrade and keep the
previous available with a simple edit of cmdline.txt

Although it's not the problem you are seeing on the Pi 5, after the
update I did at the beginning of May two of my Pi's overnight rsync
backups went from taking a few tens of seconds, to anything from minutes
to 5 hours.

I suspected the WiFi, SD cards, power supplies or even the machines themselves. After trying many combinations of those things, the common
factor was they were the only two ones on Bookworm with 512MB of RAM, everything else having 1GB or more. One was a Pi Zero 2W which started
taking minutes to backup, and the Pi 3A+ would hang for hours. I have
lot's of 512MB Pi Zero Ws but they have cameras so are still on Bullseye
so they can use RaspImjpeg with the legacy camera sack, and don't have
any problems.

The only way around it was to go back to using the old Bullseye image on
the 3A+, and dig out one of the old 1MBs Pi 2B to replace the Zero 2W.
The backups went back to taking 20 to 30 seconds. At the next round of upgrades (I only tend to do it a couple of times a year), I'll try the original configurations with Bullseye again.

---druck
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

The Natural Philosopher <tnp@invalid.invalid> wrote:

On 17/06/2025 16:08, bp@www.zefox.net wrote:

Turning wifi back on caused a momentary
dip to 5.06V which recovered within one second.

If that's a PSU problem it's mighty subtle.

It should not have dipped *at all*

Well, ten millivolts isn't much and nothing's perfect.

Have you got a storage scope you can put on it?

Yes, but it's an Owon USB 'scope so a bit of a circus to hook
up. When I get the present crop of alligators fed I'll give it
a try. For now wired networking is good enough.

My guess - without massive confidence - is that there is a momentary brownout of a lot less than 5.06V.

A very reasonable suspicion.

Thanks for writing!

bob prohaska

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

bp@www.zefox.net wrote:

The Natural Philosopher <tnp@invalid.invalid> wrote:

On 17/06/2025 16:08, bp@www.zefox.net wrote:

Turning wifi back on caused a momentary
dip to 5.06V which recovered within one second.

If that's a PSU problem it's mighty subtle.

It should not have dipped *at all*

Well, ten millivolts isn't much and nothing's perfect.

Have you got a storage scope you can put on it?

Yes, but it's an Owon USB 'scope so a bit of a circus to hook
up. When I get the present crop of alligators fed I'll give it
a try. For now wired networking is good enough.

My guess - without massive confidence - is that there is a momentary
brownout of a lot less than 5.06V.

A very reasonable suspicion.

And, a good call.

Found a spare usb-c power supply and tried it. 5.45 volts on the GPIO
pins. That seems kinda high, but the Pi5 seems happy with it. It does
grumble something about usb_max_current_limits during the boot output
but doesn't put the info in dmesg, so it's hard to act on it. Something
about messing with config.txt, but after all the trouble of the past
few days I'm just glad it boots.

Still having trouble getting dual monitors working, but that's a subject
for a new thread.

Thanks very much for writing!

bob prohaska

--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On Tue, 17 Jun 2025 21:30:31 +0100, druck wrote:

I've had far more problems with Bookworm since it moved to Network
Manager than with Bullseye and dhcpcd.

If its role is to be a server box, then configure it with a static IP and other network settings, and get rid of Network Manager, dhcpcd and all the rest of that client-side stuff. Much less trouble that way.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 2025-06-18, Lawrence D'Oliveiro <ldo@nz.invalid> wrote:

On Tue, 17 Jun 2025 21:30:31 +0100, druck wrote:

I've had far more problems with Bookworm since it moved to Network
Manager than with Bullseye and dhcpcd.

If its role is to be a server box, then configure it with a static IP and other network settings, and get rid of Network Manager, dhcpcd and all the rest of that client-side stuff. Much less trouble that way.

Amen to that!
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 18/06/2025 18:05, Jim Jackson wrote:

On 2025-06-18, Lawrence D'Oliveiro <ldo@nz.invalid> wrote:

On Tue, 17 Jun 2025 21:30:31 +0100, druck wrote:

I've had far more problems with Bookworm since it moved to Network
Manager than with Bullseye and dhcpcd.

If its role is to be a server box, then configure it with a static IP and
other network settings, and get rid of Network Manager, dhcpcd and all the >> rest of that client-side stuff. Much less trouble that way.

Amen to that!

You can use network manager to do all that via the command line.
Its not hard

Its easier than trying to resurrect older style netconfigs that will get
wiped out by upgrades
--
You can get much farther with a kind word and a gun than you can with a
kind word alone.

Al Capone



--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On 18/06/2025 09:50, Lawrence D'Oliveiro wrote:

On Tue, 17 Jun 2025 21:30:31 +0100, druck wrote:

I've had far more problems with Bookworm since it moved to Network
Manager than with Bullseye and dhcpcd.

If its role is to be a server box, then configure it with a static IP and other network settings, and get rid of Network Manager, dhcpcd and all the rest of that client-side stuff. Much less trouble that way.

I do on the headless ones, and there are less problems.

---druck
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: comp.sys.raspberry-pi

On Wed, 18 Jun 2025 18:51:48 +0100, The Natural Philosopher wrote:

Its easier than trying to resurrect older style netconfigs that will get wiped out by upgrades

I have the older style /etc/network/interfaces file going back several
years on this machine I’m using now, and on my backup. Both run Debian Unstable and undergo upgrades about once a month. I’ve never had to “resurrect” that file from an upgrade; once or twice NetworkManager does come back (tied to some dependency on the GUI desktop), but I quickly
remove that and all is well.
--- Synchronet 3.21a-Linux NewsLink 1.2