Clean install of Mageia-3-beta1-x86_64-DVD.iso. Default runlevel: 3
Package Group Selection has all package groups selected except LSB and Other
Graphical Desktops + updates followed with
urpme --auto-orphans and reboot. 

$ cat /etc/sysconfig/desktop
DISPLAYMANAGER=gdm
DESKTOP=KDE4

latest initscripts rpm resolved the problem.

Status: NEW => RESOLVED
Resolution: (none) => FIXED

New scripts and problem is back.

Resolution: FIXED => (none)
Status: RESOLVED => REOPENED

Nothing actually changed regarding that in it the initscripts package so I suspect it's somewhat fluke.

Perhaps it's failing simply because the modules in (i.e. the ones in /etc/modprobe.preload.d/*) are already loaded and that particular module doesn't like being loaded twice? Perhaps those files list no-longer present modules? Perhaps this will vary from boot to boot?

As this crops up semi-regularly, I've now removed the masking of the errors from that file. It should now tell you quite explicitly which module fails (or rather whatever error modprobe reports) in the systemctl status output (as root).

This will hopefully clear up any confusion about this "service" failing and actually point out the actual problematic module.

I'll close it as fixed as I really don't think the problem is with initscripts but rather other config on the system. Of course there could be code generating modprobe.preload.d files that's faulty, but really that should be a different bug I guess.

Feel free to keep commenting here if needed and of course reopen if there still are some real bugs here.

Status: REOPENED => RESOLVED
Resolution: (none) => FIXED

BTW Colin I've seen that error too.

CC: (none) => thierry.vignaud

clean install Mageia-4-alpha2-x86_64-DVD.iso. alpha1 also had the problem.
snipped and added blank lines follows for

$ journalctl | grep fedora-loadmodules 
 fedora-loadmodules[334]: Loading modules: evdev powernow-k8 cpufreq_powersave cpufreq_conservative cpufreq_ondemand

 fedora-loadmodules[334]: modprobe: ERROR: could not insert 'powernow_k8': No such device

 systemd[1]: fedora-loadmodules.service: main process exited, code=exited, status=1/FAILURE

 systemd[1]: Unit fedora-loadmodules.service entered failed state.

$ journalctl | grep "Kernel command line:"
 kernel: Kernel command line: BOOT_IMAGE=mga4_64_a2 root=LABEL=cauldron  nokmsboot  vga=794  

Fails on two of my systems.
[wb] $ journalctl | grep DMI:
 kernel: DMI: Hewlett-Packard p6610f/2AB1 , BIOS 6.02 07/21/2010

[tb] $ journalctl | grep DMI:
 kernel: DMI: LENOVO H215/Tilapia CRB, BIOS D2KT21AUS 03/12/2010

Resolution: FIXED => (none)
Status: RESOLVED => REOPENED

Changing the RPM and title as this is misleading considering the script is just the conduit for the problem rather than the problem itself.

I suspect we should just rip out all the cpufreq related modules code in drakx... Thomas, Thierry, do you agree? They should all be autoloaded these days.

Source RPM: initscripts-9.41-2.mga3.src.rpm => drakxtools
Summary: 3_b1: /lib/systemd/fedora-loadmodules failed to start => Invalid modules chosen for preload (exposed via fedora-loadmodules script/service)
CC: (none) => tmb

Most of the time that I see fedora-loadmodules failing, it's because of invalid modules being listed in /etc/modprobe.preload.d/cpufreq (usually the first one listed in there).  However, my sister has a Dell Inspiron 531s *desktop* system, where something keeps adding "i8k" (a module for Dell laptops that doesn't usually really work even on laptops) in the file /etc/modprobe.preload, so whatever is doing that should also be disabled.

Yes, cpufreq should be autoloaded... however it does not always work.
I have been thinking of swithing them to builtin like fedora and suse 
has done to work around the failing to load.

But I need to check before if the autoloading issues has been resolved.

iirc the i8k module is done on some dmi matching, we need to review the
code to see if can make it more accurate.

(In reply to David Walser from comment #8)
> it's because of invalid modules being listed in /etc/modprobe.preload.d/cpufreq
> (usually the first one listed in there).

Yup, $ cat cpufreq
powernow-k8
cpufreq_powersave
cpufreq_conservative
cpufreq_ondemand


(In reply to Thomas Backlund from comment #9)
 
> iirc the i8k module is done on some dmi matching, we need to review the
> code to see if can make it more accurate.

[OT] I would vote for that if code is common to other applications. :)

$ lspci | grep wireless
02:00.0 Network controller: Ralink corp. RT3090 Wireless 802.11n 1T/1R PCIe

remove-unused-packages removes the ralink-firmware rpm, harddrake2 puts it back prior to hardware scan.

for firmwares I'm going to make "remove unused packages" keep them installed as we dont know when people hotplug some hw needing firmware, and since you cant
download firmware if it's needed to get your network up...

Error is not showing up on 5 RC.

Status: REOPENED => RESOLVED
Resolution: (none) => FIXED