You are not logged in.
I was updating my packages on Arch and was trying to figure out the issue as to why hypridle did not start automatically, on Hyprland. So I rebooted to check and when I pass the bootloader, for the LUKS_LVM password input, it gets stuck in failure with 2 error messages:
[FAILED] Failure to start Load/Save OS Random Seed.
[FAILED] Failure to start Load/Save OS Random Seed.
[FAILED] Failure to start Load/Save OS Random Seed.
[FAILED] Failure to start Record System Boot/Shutdown in UTMP.
[FAILED] Failure to start Load/Save OS Random Seed.
[FAILED] Failure to start Record System Boot/Shutdown in UTMP.
Then, it seems to blink and try to login, all black for a moment, then stuck in that error message until it shutdowns.
Last thing I did was update the packages, fwupdmgr-get updates and I was trying to solve the hypridle situation but that was on hyprland.conf so a WM issue, something already inside the system which I can not get into in the first place. I checked and UTMP seems to be just a journal of logins so I believe the issue is with the Load/Save part but I don't know what exactly.
Issue persists while trying to enter through a different snapshot (just says read-only so not sure if it works tbh) on snapper and with fallback initramfs. I am with the latest kernel but I don't believe it was one of the packages installed today and I had already rebooted and logged in without issues today actually. So I really don't know what triggered this nor how I can solve it.
Offline
I figured out one of if not the entire issue. Using dysk, it seems the root system suddenly ballooned to be totally full, only 66k of free space so that is what is causing the issue. I have no idea how that happened though, since it had around 20GB free, or, always, at the very least, 15 or 16. How can I best figure out what caused this sudden increase? And how to delete without causing something to malfunction? Also, would delete and make enough room be enough to make the LUKS_LVM work again or is there another issue?
Offline
Are you cleaning up your pacman caches and logfiles / journals from time to time? Otherwise no matter how large your rootfs, it will run out of space at some point.
You can use ncdu to find out where the most space is going. (Works best from a livecd, or using mount --bind / /mnt/root, then ncdu /mnt/root)
Offline
I highly doubt it, I don't recall ever doing that but it is possible. I am chrooted but I don't have ncdu installed apparently. I am trying to see with ls or du but can't really get anything useful. How do I clean those logs? WOuld that fix the issue?
Offline
You are probably right about the logs, how can I see them and, better yet, delete them? At first I thought it might be the Snapper snapshots causing the issue but they have a cleaning algorithm and never fill up that much. I guess I did not check how full the root partition was getting. I should have figured out before rebooting because, now that I think about it, one package failed to install using pacman -Syu, I just did not think it was an issue at the time and shouldn't have rebooted it. So it has nothing to do with the Hyprland or their config but only with the logs...
Offline
chrooting is counterproductive for ncdu. for chroot you need proc sys dev etc., but for ncdu you don't want it.
you should be able to install it in the livecd environment if not present by default.
Offline
So unmounting the current installation and use it only on liveCD?
Offline
Managed to check with
du -sh ./*
and cd'ing deeper and deeper, it really is the cache with ./pacman the real culprit with 24G!
Can I just
rmdir
without issues or is there a safer way? And will that automatically fix the issue with the Load/Save Random OS Seed or is there another step?
Offline
I did
pacman -Sc
to clear old packages in cache and now the directory pkg is "only" 4.6G instead of 20G so that, in theory, should solve it... however, it still shows the arch-root as being totally full using dysk and when I try to enter again, still the same error. Is it an issue with proc? Because it marks as not being able to read on chroot when I do
du -sh ./*
, for some reason...
Says that
du: cannot access '.proc/1894/
So, basically, the pkg in ./cache/pacman is now just 4G but doesn't matter because it still marks as only have 66k space left. Is there a command to sync that or is there something else wrong?
Last edited by sdsd9 (2025-05-18 16:01:40)
Offline
The problem persists, even though I managed to login but now I find another error when trying to access the luks volume.
[FAILED] Failed to start Create System Users.
[FAILED] Load/Save OS Random Seed.
Rebuild Journal catalog
Failed to start Record System Boot/Shutdown in UMTP
Rebuild Dynamic Linker Cache.
Failure to listen on GnuPG network certificate management Daemon for /etc/pacman.d/gnupg
GNuPG cryptograhipc agent and passphrase cache (access for web browsers) for /etc/pacman.d/gnupg
Failed to start Take snapper snapshot
and, once again can't login. I wasn't able to do pacman -Syu while logged in and regrettably, I thought a reboot might fix that
Offline
I was having issues even using liveCD but seems to work now. I chrooted and everything and even though there is network (checked with ping) while chrooted, I can't even do
pacman -Syu
giving me the error of
failed to synchronize all databases (unexpected error)
I wonder if it's a lock it created and I have to delete, just not sure. Something related to authentication? How do I check? I have snapper so I could do a rollback technically but I don't think that would fix it even if it works.
Offline
I was booting through snapshots, I think there is some weird, badly configuration there so booting in read mode and says that the snapshots is full (even though the disk has plenty of space) so that is probably part of the problem. Without liveCD and editing the boot entries and changing the params back to
subvolid=/
I manage to access some of it so it does not yield the exact same errors but still the Load/Save OS Random Seed. Furthermore, it says that it cannot mount the efi and prompts to see the root password. I input that once again and get the terminal [root@username] and I can see the files so I suppose that's an improvement there. Gonna read the btrfs but I am still confused and wondering how to go about it, seems like it ought to be a simple alteration in the boot but I am failing to understand it. Also, the files seem themselves to be intact and not old snapshots, I checked alterations I had done just over a day ago in some of them and, other than not being able to save on VIM, probably due to the services not starting properly, the data is there.
What should be the next step to try?
Offline