Boot time has increased after 2022-12-20 Stable update (due to userspace)

Support
22

Thanks. I did keep some records, but they are probably not that useful now. As you can see boot time has almost doubled since gnome 42 updates. probably should system reinstall again.


##### START UP TIME ########
before gnome42	
	systemd-analyze
Startup finished in 3.110s (firmware) + 1.172s (loader) + 1.739s (kernel) + 4.612s (userspace) = 10.635s 
graphical.target reached after 4.611s in userspace

after
	systemd-analyze
Startup finished in 3.140s (firmware) + 852ms (loader) + 1.898s (kernel) + 6.290s (userspace) = 12.182s 
graphical.target reached after 6.289s in userspace

~~~~ 2022.06.13
Gnome 42.2 + kernel 5.18

systemd-analyze                                                  ✔ 
Startup finished in 3.114s (firmware) + 1.781s (loader) + 2.012s (kernel) + 6.614s (userspace) = 13.523s 
graphical.target reached after 6.614s in userspace.
23

Probably then makes sense to ask OP about used edition/DE. I could not on Xfce in the end confirm, although I thought I saw the same thing…

24

This is not a problem of the graphic interface. I think it’s a problem between the kernel and systemd. I use KDE

25

I use Xfce.

26

Do you have a copy of what your /etc/mkinitcpio.conf:HOOKS= line looked like before as compared to now?

That is. The 24-12 stable update just blew up my system by having me boot with an initramfs without mdadm_udev in there whereas my rootfs is on mdadm raid0. With no mdadm installed onto the initramfs I had to scramble a bit to get it back up.

Now, I’m relatively positive I’ll have screwed that up myself; new user and I do remember sudo pacdiff post 20-12 making me change /etc/mkinitcpio.conf. I expect I mismerged due to inexperience – but that is at the same time also somewhat unexpected since I’m due to having needed to custom set that up during install pretty much aware of needing that mdadm hook in there (and how I hadn’t noticed before beats me since certainly that was before the 24-12 update).

So just maybe I didn’t in fact screw up myself and you saw your HOOKS= adjusted similarly with in your case new and slow steps inserted…

For comparison, my current fairly minimal one is

HOOKS=(base udev autodetect modconf block mdadm_udev filesystems)

Make sure to run sudo mkinitcpio -P after changing it.

[EDIT] Yes, noticing on review again that your issue seems to be in userspace – but I’ll leave this up anyway; an initramfs difference as to e.g. KMS would supposedly have an effect on userspace init time after all.

27

HOOKS=(systemd autodetect modconf kms block filesystems keyboard sd-vconsole fsck)

28

AFAIA you should in fact “always” have those HOOKS start with “base” – but in any case the here important thing is it currently being the same or not as before. I.e., I also saw kms in mine even though I was pretty sure it wasn’t previously. Depending on your graphics hardware that could make a difference.

29

My HOOKS

HOOKS=(systemd autodetect modconf keyboard keymap block filesystems)

30

Very unlikely then that it has in your case anything to do with things. I should and will stay out; hoped my own little FUBAR as to losing mdadm_udev from there would at least point to somewhere useful but it likely/seemingly does not (although, post-edit, again no “base”?)

31

From

https://wiki.archlinux.org/title/Mkinitcpio

Sets up all initial directories and installs base utilities and libraries. Always keep this hook as the first hook unless you know what you are doing, as it provides critical busybox init when not using systemd hook.
Optional when using the systemd hook as it only provides a busybox recovery shell.

32

I see…