Getting systemd console message before reboot/shutdown

Witcher619 · April 12, 2025, 8:45pm

Hi everyone,
All of a sudden I have started to receive a weird momentary console message from systemd about something related to watchdogs? (I have no idea what this is). The message says:
#systemd-shutdown[1]: Failed to set watchdog hardware timeout to 1.
Omin: Invalid argument#
See the screenshot attached as well.
This message appears just for a fraction of a second and does not impede the use of the laptop in any way, but it’s visually unappealing and annoying, since I don’t understand why it appears and how to disable it.
I tried browsing through the web, but found nothing similar, or at least any possible solutions to it, so perhaps you people could help me figure it out?
Not sure which info you might need from me, but my boot loader is rEFInd, I don’t have either systemd or grub.
Thanks for your help in advance.

Cphusion · April 12, 2025, 9:13pm

If it ain’t broke don’t fix it? Not all error messages need to be fixed. If it bothers you that much you can probably try this. I can’t make any promises though.

Witcher619 · April 13, 2025, 10:08am

I mean, you are right, but first and foremost:

Why does this message even appear?
It’s ugly and unneeded, hence I want it to be removed
The linked article, from my understanding, instructs how to enable this watchdog thing, but I don’t need it at all. Can I just delete this watchdog? Or somehow silence it completely?

cactux · April 13, 2025, 10:20am

Do you have nowatchdog in your kernel command line?

Cphusion · April 13, 2025, 10:46am

First time hearing of watching do, it seems to be there to recover your system when it detects malfunctions. It seems to have to do something with watchdog hardware, it was fixed from an issue I can find.

github.com/systemd/systemd

Failure to set watchdog timeout error is unnecessarily ambiguous

opened 09:30PM - 06 Mar 24 UTC

closed 03:53PM - 04 Apr 24 UTC

vcaputo

bug 🐛 needs-better-log-message watchdog

### systemd version the issue has been seen with unspecified ### Used dist…ribution unspecified ### Linux kernel version used unspecified ### CPU architectures issue was seen on arm ### Component watchdog ### Expected behaviour you didn't see Clear message characterizing the failure as a watchdog-specific timeout issue. ### Unexpected behaviour you saw ``` [123456.1234] systemd-shutdown[1]: Failed to set timeout to 10min: Invalid argument ``` ### Steps to reproduce the problem Have an RestartWatchdogSec (default is 10mins) unsupported by the underlying hardware watchdog, reboot/shutdown. This is all fairly trivial, just see https://github.com/systemd/systemd/blob/main/src/shared/watchdog.c#L277 It's a one-liner to improve, unfortunately I tripped over this at $dayjob and don't yet have authorization to contribute even trivial changes upstream AIUI. Even filling out the submission form is kind of tricky, for now, sorry folks.

But for some reason it’s appearing for you. That’s why I thought it might be worth a try for you to set it to what’s recommended in the wiki to see if doesn’t appear then, if the time limit was lower. Error messages aren’t meant to win a beauty pageant.

To turn the hardlockup detector off, use:

/etc/sysctl.d/disable_watchdog.conf

kernel.nmi_watchdog = 0

or add nmi_watchdog=0 to the kernel line.

Alternatively add nowatchdog to the kernel line to disable both hard and soft lockup detectors. See [3]

Witcher619 · April 14, 2025, 7:48pm

Hey,
Thanks for getting back.
I have the “nowatchdog” parameter specified in my kernel parameters. I have also added the “nmi_watchdog=0” parameter in there, just in case, as the issue was still persistent with just the"nowatchdog" parameter.
So I’m kind of stuck here, as these basic kernel parameters don’t work, unfortunately.

cactux · April 14, 2025, 7:51pm

 lsmod | grep iTCO_wdt

Do you get any output?

Witcher619 · April 17, 2025, 10:18am

Hey,
Sorry for a slightly delayed reply, was a bit busy.
This command does not give any output.
Also, I have confirmed through the “journalctl -b | grep watchdog” command that the only watchdog references I get are from my kernel parameters. But nevertheless, I keep getting this “failed to set watchdog hardware timer” error.
Maybe I really need to try and set it in the first place, I don’t know.
But why does this error still keep appearing if I have allegedly turned off everything watchdog related?
Thanks a lot in advance

Witcher619 · April 18, 2025, 11:05am

Funnily enough, I was able to fix this issue with the help of ChatGBT.
Having described the issue to it, ChatGBP suggested the following kernel parameter: systemd.watchdog_device=none
Why it is funny is because nowhere have I found at least any vague mention towards this parameter in my hours of searching the web for any clues on how to tackle this error. Literally 0 references, only something generic like “nowatchdog” or “nmi_watchdog=0”, but ChatGBP managed to solve this issue that I’d been trying to solve for days with literally the 1st reply…
If anyone’s interested why this has worked, here’s a reply from ChatGBP:

Parameter	What it Affects	Why it Fails Here
`nowatchdog`	Kernel’s internal soft lockup watchdog	Not related to systemd watchdog
`nmi_watchdog=0`	Disables NMI watchdog used for kernel debugging	Not systemd-related
`watchdog_timer=0`	Some driver-specific watchdog settings	Doesn’t affect systemd or userland
Only systemd cares about `systemd.watchdog_device`
So if your error message is from systemd — like `Failed to set hardware watchdog timer` — then none of the above will touch it, because that’s happening after the kernel has booted and systemd is taking over.
God praise ChatGBP

system · April 20, 2025, 11:06am

This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.