Udev rules do not work sometimes

gxhary · 2026-01-26 03:03:19

I have a python script that I ran as a cron job to notify me when my battery was about to die. I wanted to change that cron job into a udev rule, but I cannot get it to work at all.

The main issue I faced is: It sometimes works and some other times it doesn't and I don't know how to replicate this issue.
How can I fix this?
Is there a way to manually trigger the udev event rather than waiting for my battery to charge then die and repeat? This would at least help me replicate the issue and figure out whats wrong

This is my udev rules file

# Rule to send a notification when battery is below 20% 
# This one sometimes work
SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", ATTR{capacity}=="[0-1][0-9]", ENV{DISPLAY}=":0", RUN+="/usr/bin/su gxhary -c '/home/gxhary/.config/scripts/battery_watcher.py battery-low'"

# these two do work at exact values
SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", ATTR{capacity}=="20", ENV{DISPLAY}=":0", RUN+="/usr/bin/su gxhary -c '/home/gxhary/.config/scripts/battery_watcher.py battery-low'"

# Rule to send a notification when battery is fully charged 
SUBSYSTEM=="power_supply", ATTR{status}=="Charging", ATTR{capacity}=="100", ENV{DISPLAY}=":0", RUN+="/usr/bin/su gxhary -c '/home/gxhary/.config/scripts/battery_watcher.py battery-charged'"

Another issues I faced is I can't get it to work at 20% or 100%, (exact values) I can only get it to work (when it does work) at ranges.

Thanks in advance

Last edited by gxhary (2026-01-26 03:04:16)

seth · 2026-01-26 13:17:12

See the note in https://wiki.archlinux.org/title/Laptop#udev - if the battery only emits discharge events < 5% but dies at 10% you'll never see those and polling will unfortunately remain your only option

gxhary · 2026-01-26 15:22:14

seth wrote:

See the note in https://wiki.archlinux.org/title/Laptop#udev - if the battery only emits discharge events < 5% but dies at 10% you'll never see those and polling will unfortunately remain your only option

No that's the thing, it occasionally works, but it is not consistent. So my battery does send those events, but the udev rules arent always properly triggered

seth · 2026-01-26 15:32:24

Does it really ever trigger on capacity == "09" (rather than just "9")?
Did you test whether the udev rule fires or whether your script does the expected thing?
Try a rule that does sth. like "/usr/bin/touch /tmp/udev.fired" to track the invocation.

gxhary · 2026-01-26 17:49:40

seth wrote:

Does it really ever trigger on capacity == "09" (rather than just "9")?
Did you test whether the udev rule fires or whether your script does the expected thing?
Try a rule that does sth. like "/usr/bin/touch /tmp/udev.fired" to track the invocation.

When my battery is 9% it does trigger, so my guess is yes it triggers at 09 because that's the only rule I have ([0-1][0-9]).
It does not work when I try to match a single number like 20 or 100.
My script works as expected, I ran it outside the udev rule and it works fine.

I am also using the same exact script as a udev rule for when plugging and unplugging the charger and it works fine.
The problem is not my script.

The problem is udev rules for when the battery is discharging occasionally works.
I do not understand why and I cannot replicate it.

- Is there a way to trigger the udev rules manually? I would be able to debug it on my own.
- If not then how can I properly test it?
- Is my syntax correct for the single number matching? I couldn't something helpful for it. Because apparently

ATTR{capacity}=="20"

is not correct

seth · 2026-01-26 20:55:15

It does not work when I try to match a single number like 20 or 100.

Specific percentage might be skipped?

My script works as expected, I ran it outside the udev rule and it works fine.

So it's just useless…
Track what's actually going on.

SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", RUN+="/tmp/chargelog.sh '$attr{capacity}'"

/tmp/chargelog.sh

#!/bin/sh
echo $1 >> /tmp/charge.log

will hopefully do this (I don't get discharge events > 8% or so and am currently at 98& w/ a remaining time > 10h

gxhary · 2026-01-27 00:58:35

seth wrote:

It does not work when I try to match a single number like 20 or 100.
Specific percentage might be skipped?
My script works as expected, I ran it outside the udev rule and it works fine.
So it's just useless…
Track what's actually going on.
SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", RUN+="/tmp/chargelog.sh '$attr{capacity}'"
/tmp/chargelog.sh
#!/bin/sh
echo $1 >> /tmp/charge.log
will hopefully do this (I don't get discharge events > 8% or so and am currently at 98& w/ a remaining time > 10h

Thank you I will do this and report back, hopefully this helps me fix it

gxhary · 2026-01-27 16:05:44

seth wrote:

It does not work when I try to match a single number like 20 or 100.
Specific percentage might be skipped?
My script works as expected, I ran it outside the udev rule and it works fine.
So it's just useless…
Track what's actually going on.
SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", RUN+="/tmp/chargelog.sh '$attr{capacity}'"
/tmp/chargelog.sh
#!/bin/sh
echo $1 >> /tmp/charge.log
will hopefully do this (I don't get discharge events > 8% or so and am currently at 98& w/ a remaining time > 10h

I tried it and I wasn't getting discharge events at all.
Unless I can get it to send events, I guess I will just stick with a cron job
Thank you for your help

seth · 2026-01-27 16:26:56

Then how do you previously receive events?
Forgot to xhmod ugo+x /tmp/chargelog.sh ?

SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", RUN+="/usr/bin/touch /tmp/discharge.proof"

mpan · 2026-01-27 19:05:17

Gxhary, make sure you’re actually matching the right thing. More precisely: that you’re selecting the most specific device.

Events for a single device usually have multiple matches at different levels of specificity. The less specific ones may be missing some attributes you’re testing against. `udevadm monitor` and `udevadm info --attribute-walk` are your friends.

On the part regarding matching exact values: do not. This is nothing with udev, it’s sampling 101. Intermediate values may be missing in a sample of a varying function. Always test for ranges. It’s going to be less tedious, if that’s being done in the script itself. You have the freedom of using more complex logic in a script.

seth · 2026-01-27 21:40:56

Fwwi, #6 works for me (at least I get a single discharge when unplugging the power and it logs 100, again - don't expect further messages > 8% or so)

gxhary · Yesterday 02:06:10

seth wrote:

Then how do you previously receive events?
Forgot to xhmod ugo+x /tmp/chargelog.sh ?
SUBSYSTEM=="power_supply", ATTR{status}=="Discharging", RUN+="/usr/bin/touch /tmp/discharge.proof"

That's what i am saying, it's inconsistent.
What exactly happened is as follows:
1. I set the rule as you said
2. I tested the script with another rule that I know works and can easily replicate
3. I used my laptop normally until it died
4. I frequently checked the log file and saw nothing
5. Interestingly, when I closed the laptops lid to go from the living room to my bedroom, I noticed it did log, somehow that means that I get events when my laptop is idle?
6. I only got events when I closed my lid and opened it back, but not when I was normally using it

mpan wrote:

`udevadm monitor` and `udevadm info --attribute-walk` are your friends.

does that mean i can run `udevadm monitor` and see the discharge events when/if they happen?

Update: I ran `udevadm monitor` till my laptop died and still got nothing

Last edited by gxhary (Yesterday 03:21:28)

seth · Yesterday 09:00:42

At what percentage does your battery typically die?

mpan · Yesterday 14:02:35

gxhary wrote:

mpan wrote:
`udevadm monitor` and `udevadm info --attribute-walk` are your friends.
does that mean i can run `udevadm monitor` and see the discharge events when/if they happen?

It’s going to report all events that udev sees, reporting the device path for each. If events happen in relation to your laptop battery, the associated device should appear there. You can then analyze, what attributes and at which level of specificity are available.

gxhary wrote:

Update: I ran `udevadm monitor` till my laptop died and still got nothing

Wait? Absolutely no output?

gxhary · Yesterday 16:39:50

seth wrote:

At what percentage does your battery typically die?

Normally at like 1% or 0%

mpan wrote:

Wait? Absolutely no output?

No output related to the battery, I got some for other things like changing my brightness and plugging/unplugging my charger

seth · Yesterday 20:44:27

So it's not clear whether the rule has actually ever worked or the script invocation was external?
=> https://archlinux.org/packages/extra/x86_64/batsignal/

gxhary · Yesterday 21:44:22

seth wrote:

So it's not clear whether the rule has actually ever worked or the script invocation was external?

I think I figured out what is happening. I think I only get discharge events when I unplug my charger.
I been trying to figure it out and I realized I only get a notification when I unplug my charger.
I was able to get it consistently to work by plugging in the charger, waiting a few seconds, then unplugging it.
I would consistently get the rule to work.
I guess this means that my batter doesn't send discharge events at certain percentages, rather it sends it when it starts discharging.

This is the udevadm monitor output when I was able to replicate it

[gxhary@archlinux ~]$ udevadm monitor
monitor will print the received events for:
UDEV - the event which udev sends out after rule processing
KERNEL - the kernel uevent

KERNEL[55247.213898] change   /devices/platform/ACPI0003:00/power_supply/AC (power_supply)
KERNEL[55247.254560] change   /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0A:00/power_supply/BAT0 (power_supply)
UDEV  [55247.263351] change   /devices/platform/ACPI0003:00/power_supply/AC (power_supply)
UDEV  [55248.329342] change   /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0A:00/power_supply/BAT0 (power_supply)

Thank you for the help

Edit: Also Thank you for linking the daemon, my plan was to get the udev rules to work, my backup was to make my own daemon. I'll take some inspiration from it and make my own in rust lol, thanks

Last edited by gxhary (Yesterday 21:49:38)

Arch Linux

#1 2026-01-26 03:03:19

Udev rules do not work sometimes

#2 2026-01-26 13:17:12

Re: Udev rules do not work sometimes

#3 2026-01-26 15:22:14

Re: Udev rules do not work sometimes

#4 2026-01-26 15:32:24

Re: Udev rules do not work sometimes

#5 2026-01-26 17:49:40

Re: Udev rules do not work sometimes

#6 2026-01-26 20:55:15

Re: Udev rules do not work sometimes

#7 2026-01-27 00:58:35

Re: Udev rules do not work sometimes

#8 2026-01-27 16:05:44

Re: Udev rules do not work sometimes

#9 2026-01-27 16:26:56

Re: Udev rules do not work sometimes

#10 2026-01-27 19:05:17

Re: Udev rules do not work sometimes

#11 2026-01-27 21:40:56

Re: Udev rules do not work sometimes

#12 Yesterday 02:06:10

Re: Udev rules do not work sometimes

#13 Yesterday 09:00:42

Re: Udev rules do not work sometimes

#14 Yesterday 14:02:35

Re: Udev rules do not work sometimes

#15 Yesterday 16:39:50

Re: Udev rules do not work sometimes

#16 Yesterday 20:44:27

Re: Udev rules do not work sometimes

#17 Yesterday 21:44:22

Re: Udev rules do not work sometimes

Board footer