udev-worker log entries after new kernel...?

[G3nt00]

Hi,

I really don't know where to ask this, so starting here. I unstalled new gentoo-sources & compiled genkernel yesterday, now running 6.1.31.

I noticed some odd errors in the log I haven't seen before

Code: Select all

Jun  2 04:45:32 pc (udev-worker)[1453]: could not read from '/sys/module/pcc_cpufreq/initstate': No such device
Jun  2 04:48:50 pc (udev-worker)[2092]: sunrpc: Process '/sbin/sysctl -q --pattern ^sunrpc --system' failed with exit code 1.
Jun  2 04:48:50 pc (udev-worker)[2092]: lockd: Process '/sbin/sysctl -q --pattern ^fs.nfs.n[sl]m --system' failed with exit code 1.
Jun  2 04:48:50 pc (udev-worker)[2098]: nfs: Process '/sbin/sysctl -q --pattern ^fs.nfs --system' failed with exit code 1.
Jun  2 04:48:50 pc (udev-worker)[2092]: nfsv4: Process '/sbin/sysctl -q --pattern ^fs.nfs.(nfs_callback_tcpport|idmap_cache_timeout) --system' failed with exit code 1.

What is that all about? I tried google, but didn't find much at all... Any guidance please?

Edit: More investigation shows that no udev rules exist in /etc/udev/rules.d/... Should there not be rules for the services above?

Edit2:
When checking for sys-fs/udev I found that uninstalled and masked... So added it to /etc/portage/package.unmask, but then got loads of errors;

Code: Select all

# emerge --ask sys-fs/udev

These are the packages that would be merged, in order:

Calculating dependencies... done!
Dependency resolution took 3.22 s.

[ebuild  N     ] sys-apps/systemd-utils-252.9  USE="acl kmod (split-usr) tmpfiles udev -boot (-selinux) -sysusers -test" ABI_X86="(64) -32 (-x32)" 
[ebuild  N    #] sys-fs/udev-250  ABI_X86="(64) -32 (-x32)" 
[blocks B      ] sys-apps/systemd ("sys-apps/systemd" is soft blocking sys-apps/systemd-utils-252.9)
[blocks B      ] sys-apps/gentoo-systemd-integration ("sys-apps/gentoo-systemd-integration" is soft blocking sys-apps/systemd-utils-252.9)
[blocks B      ] sys-fs/udev ("sys-fs/udev" is soft blocking sys-apps/systemd-253.3-r1, sys-apps/gentoo-systemd-integration-9)

 * Error: The above package list contains packages which cannot be
 * installed at the same time on the same system.

Well, I do understand the least lines, but why aren't udev installed and why the log-errors? What should I do to resolving this please?

Thanks

[NeddySeagoon]

G3nt00,

Code: Select all

[ebuild  N     ] sys-apps/systemd-utils-252.9  USE="acl kmod (split-usr) tmpfiles udev -boot (-selinux) -sysusers -test" ABI_X86="(64) -32 (-x32)"
[ebuild  N    #] sys-fs/udev-250  ABI_X86="(64) -32 (-x32)" 

If you are an OpenRC user one of those two packages should be installed but they are both flagged N for New, so neither is at the moment.
Further, sys-fs/udev is masked which is what the # means.

Three of the systemd 'offcuts' that were packaged separately for OpenRC users have been merged into sys-apps/systemd-utils for ease of maintainance by both devs and users as all three parts needed to be updated at the same time.

eudev users need to use sys-apps/systemd-utils with USE=-udev on sys-apps/systemd-utils only as eudev uses some of the same filenames as udev, wherever udev comes from.

None of this applies to systemd. Systemd provides udev too, as its an integral part of systemd.

If you are an OpenRC user, you either have eudev installed or udev is missing from your system entirely. That would explain the lack of udev rules.
You need sys-apps/systemd-utils to fix that.

If you are a systemd user, its something else. I can't help there as I have no experience with systemd.

[G3nt00]

G3nt00,

If you are a systemd user, its something else. I can't help there as I have no experience with systemd.

Hi, thanks for responding. Unfortunately I am using Systemd. I know Gentoo have its legacy in initrc which later became OpenRC, so perhaps that is what I should have gone with when installing. From docs it sounded to me as if Systemd was the future, but if that is not the Gentoo strategy, I have misread that.

If anyone can help where to start to look for a possible problem I would much appreciate it, and understand how much "needle in a haystack" kind of thing it really may be, but hope it makes sense to someone.

I do have a ZloneZilla backup these days, but most likely taken after this happened, so I look at it as a "back to working at all" more than "restore to mint".

Thanks

[hdcg]

Hi G3nt00,

looking at the logs you posted, I am not sure your issue is init system related.
They are all caused by /sys related errors.

Are you sure the modules of your newly installed kernel are installed/loaded properly? What's the output of

Code: Select all

lsmod

?

And if you have your old kernel available still as a bootable backup whats the outputs of

Code: Select all

ls -lA /sys/module/pcc_cpufreq
ls -lA /sys/fs

(for new kernel as well as old kernel).

Best Regards,
Holger

[G3nt00]

Hi G3nt00,

looking at the logs you posted, I am not sure your issue is init system related.
They are all caused by /sys related errors.

Are you sure the modules of your newly installed kernel are installed/loaded properly? What's the output of

Code: Select all

lsmod

?

And if you have your old kernel available still as a bootable backup whats the outputs of

Code: Select all

ls -lA /sys/module/pcc_cpufreq
ls -lA /sys/fs

(for new kernel as well as old kernel).

Best Regards,
Holger

Hi, lsmod shows me a huge list of loaded modules, so seems ok. I will post if it helps. For the running kernel the file '/sys/module/pcc_cpufreq' does not exist, and issuing 'ls -lA /sys/fs' yields

Code: Select all

ls -lA /sys/fs
total 0
drwx-----T  2 root root 0 Jun  3 08:10 bpf
dr-xr-xr-x 11 root root 0 Jun  3 08:10 cgroup
drwxr-xr-x  4 root root 0 Jun  3 08:10 ext4
drwxr-xr-x  3 root root 0 Jun  3 08:10 fuse
drwxr-xr-x  3 root root 0 Jun  3 08:10 nfs
drwxr-x---  2 root root 0 Jun  3 08:10 pstore
dr-xr-xr-x  2 root root 0 Jun  3 08:57 resctrl
dr-xr-xr-x  2 root root 0 Jun  3 08:10 selinux

I will now reboot into the old kernel and see what that looks like
Edit: It was the same result as above with the previous kernel, 6.1.28

I also tried issuing the second command of the initial list

Code: Select all

sysctl -q --pattern ^sunrpc --system
sysctl: cannot open "/usr/lib/sysctl.d/99-sysctl.conf": No such file or directory

A 'ls -la /etc/sysctl.d/' shows it is completely empty...

[NeddySeagoon]

G3nt00,

Gentoo does not have a strategy at all.

Its a loose collection of devs pursuing their own interests. There is cooperation where one part impacts another, with the final arbiter being the council.
That does not make for a strategy.

It has both systemd and OpenRC because the devs want to support both.
Systemd and OpenRC are different. Its rather like editor wars.

In the interests of full disclosure, systemd does not solve any problems that I appear to have, so I have no need to invest time in learning it.

[G3nt00]

G3nt00,

Gentoo does not have a strategy at all.

Its a loose collection of devs pursuing their own interests. There is cooperation where one part impacts another, with the final arbiter being the council.
That does not make for a strategy.

It has both systemd and OpenRC because the devs want to support both.
Systemd and OpenRC are different. Its rather like editor wars.

In the interests of full disclosure, systemd does not solve any problems that I appear to have, so I have no need to invest time in learning it.

Oh, that was a bit of news still.. I knew from before that Gentoo was not to be considered a distro, but I thought there were some sort of direction or strategy, but good to know, thanks.

I used intrc way back and found it a lot easier and straight forward than later Systemd. The distros I was currently into when Systemd was introduced jumped on that train and I with it. Perhaps time to re-evaluate. I automatically assumed Systemd would bring more, but perhaps also complexity. I miss the simplicity of early Linux. Everything has been so intertwined and complex to handle these days, maybe it has too.

I know by now that one should not jump between Systemd and OpenRC without cause, and that it may cause other issues, but if I were to investigate how to convert my system into a pure OpenRC one, where would I start? And please tell me if a conversion usually has good success rates, or if a reinstall is the better choice (but suppose not)

[NeddySeagoon]

G3nt00,

I have never done the conversion.

Early in the systemd story it was possible to install both systemd and openrc and pick one at boot time depending on the init= kernel parameter bassed by the boot loader.
I don't know if that is still possible.

[G3nt00]

G3nt00,

I have never done the conversion.

Early in the systemd story it was possible to install both systemd and openrc and pick one at boot time depending on the init= kernel parameter bassed by the boot loader.
I don't know if that is still possible.

Well, that method sounds like a good plan if it still works like that. I will try to find out.

As of now I had to restore a backup, so have finally tried it in an IRL situation and worked great, alas I lost a few things but nothing important.

I think I will do a bold move and select a new profile and see how that goes. If all else fails, I can restore up to this point at least

Edit: Well, not too surprisingly - NOT a good idea (I knew it, but sometimes luck is with you). From googling, it seems some have had success converting, their solutions have not worked for me so far, so suppose a reinstall is better for me. I have most (hope all) documented as I want it now, apart from the used profile then.

[NeddySeagoon]

G3nt00,

It should be fixable, there is no need for a reinstall.
Unfortunately, I can't help much.

[G3nt00]

G3nt00,

It should be fixable, there is no need for a reinstall.
Unfortunately, I can't help much.

I have now reinstalled after all, seemed the sensible thing to do after all. I understand it probably was overkill to do that, but a clean slate is always nice, and perhaps faster. Now almost fully up and running again, I find that localization don't work as expected. I did follow the handbook, this time for OpenRC, and did not get any errors, but

in /etc/X11 there is no 'xorg.conf.d' for example. It is instead found in '/usr/share/X11/'.

Edit:I found that the dir above should just be created if missing. Instead I made a symlink, is that not a good thing? It does still not work despite I created a file:

Code: Select all

cat /etc/X11/xorg.conf.d/30-keyboard.conf 
Section "InputClass"
    Identifier "keyboard-all"
    Driver "evdev"
    Option "XkbLayout" "sv-latin1"
    Option "XkbVariant" "altgr-intl,abnt2"
    MatchIsKeyboard "on"
EndSection

It should be simple enough I assumed, but perhaps not being read...?

Doing 'setxkbmap -query'

Code: Select all

rules:      evdev
model:      pc105
layout:     sv-latin1
variant:    altgr-intl,abnt2

shows what I configured, but is still not working in Konsole or any apps when running X11... What do I miss? I even installed 'x11-lib/libsxklavier', that apparently modern applications use according to docs, but no dice...

Edit2:
I managed to get keyboard layout working, but still don't understand. I had to configure this X11 (Keyboard - System settings), despite already configured on a system level through config files... Why aren't they being used?

Thanks

[NeddySeagoon]

G3nt00,

The symlink is not required.
/usr/share/X11/ is managed by portage and Xorf looks there for configuration options first.
Any changes you make will be lost at upgrade time.

Your overrides go in /etc/X11/... either xorg.conf which is the old way or in xorg.conf.d/ which is the new way.

[G3nt00]

G3nt00,

The symlink is not required.
/usr/share/X11/ is managed by portage and Xorf looks there for configuration options first.
Any changes you make will be lost at upgrade time.

Your overrides go in /etc/X11/... either xorg.conf which is the old way or in xorg.conf.d/ which is the new way.

Thank you, removal of the link I will then