You are not logged in.
@aluma, "amprolla" is not involved at all in the delivery of packages. "amprolla" runs a program that reads the meta files from the debian and devuan-only repositories and "merges" them by creating the merged meta files which it then distributes to pkgmaster.devuan.org.
pkgmaster.devuan.org serves packages. It's repository (both the devuan-only and the merged parts) is also mirrored to several other servers, and the domain name "deb.devuan.org" is used to obtain all IP addresses for all those (inkluding pkgmaster.devuan.org).
Of course, DNS resolution is a separate subsystem that mostly doesn't involve devuan servers except for eventually and intemittently obtaining that resolution list for deb.devaun.org, which nameservers here and there digests into their cache for the in-between-times resolution service.
All in all, the most problems arise in the DNS resolution sub system; it almost never leads back to hickups with the cllient networking software. Usually due to the DNS subsystem being in transition with intermediate servers being updated. On occasion there are also transient inconcistencies in downloads due to the meta files have been updated on the actual server queried for those before the package pool of the actual server queried for those.
I remain amazed about FOSS people just ignoring that github is owned by Microsoft.
There are so many other places to use rather than supporting those guys.
@shimarin, thanks.
The initial step of that installer is a dialog with a menu of 8 options plus 'help'. You want me to guess that you used option '1 Install' (that's ok) and I can probably likewise guess from your outline for the handful of dialogs following; in particular that in the task selection dialog you unticked everything except the bottom one for standard utilities.
Eventually reboot into the installed system, which presents the login prompt on vt1.
So what's next? Login as root or user? And perhaps you can include the output of
grep ^Commandline /var/log/apt/history.log | grep -n ''up to the point where you first ran into the Xorg issue.
Yes. You should make sure daedalus-security is in your sources, and upgrade.
1. Install media. This would be the filename of the iso file that you used to install devuan from. Perhaps even the URL you used. If you have done "installed devuan" some other way than using an installation media, then you would need to descritbe that too. Only so that I can do the same.
2. When installing with the installer ISO it takes you through a series of dialogs. I was asking that you tell me what to do so I can get the same installation done.
3. The boot command line is found in /var/log/syslog as a line with the text "Kernel command line:" in. It would tell me the specific kernel arguments used when the system starts. Presumably you would have "the default" from your manner of installation, so this point is merely a way for me to be ensured that my test system from points 1 and 2 runs in the same way as yours.
So don't worry about point 3, but please make sure you are thorough about points 1 and 2.
Could you tell what "fresh install of Devuan" means?
Which media did you use?
Which installer choices did you make?
What's your boot command line?
Would be good if I can replicate the issue; there have been similar Xorg crash reports before and possibly there's a common cause.
Did you install xserver-xorg-video-* package(s)?
Who is running startx; root or non-root?
The short story is "yes, install seatd; log out and log in". That will make you run Xorg while avoiding the buggy code path of logind.
The long story is longer but who cares?
How did you install? Which desktop environment did you choose? How have you got so many packages from "unknown" (especially both the nouveau and nvidia modules)?
You might try to reinstall libc6; version 2.31 is from chimaera:
apt-get install --reinstall libc6But be sure to have a rescue boot at hand, just in case, because it's a rather fundamental system transition. Otoh it should have happened on the upgrade... did you "dist-upgrade" at all?
Ok. Remove the link /usr/lib/x86_64-linux-gnu/libc.so.6 and run (as root)
# ldconfigwhich usually does good things to the libraries. You might also need to set a new link (as root)
# ln -sTf /lib/x86_64-linux-gnu/libc.so.6 /usr/lib/x86_64-linux-gnu/libc.so.6to sort out that dangling one.
I would think this all has to do with system differences between bullseye and daedalus (bookworm), and yet it's peculiar that only expr brings a problem.
Ok. Some of those "libraries" are actually "ld scripts" rather than actual libraries; though I don't have a clear idea how it works... try:
$ find /usr/lib* /lib* -name libc.so\* | xargs ls -l to get an overview of the "mess" ![]()
It's still peculiar that only expr is affected.
Peculiar. Try running
$ LD_DEBUG=all exprwhich should dump a haystack of goodies from the ELF loader. At a glance the problem appears to be with the ELF loader but I would then have expected it to concern all programs.
The haystack should include lines like
17051: checking for version `GLIBC_2.34' in file /lib/x86_64-linux-gnu/libc.so.6 .... Those indicate the particular ABI versions that the program requires. The hope is that there is some clue for the problem towards the end of it.
It looks to me that the general commercial drive is to make Linux a platform for "Apps", probably so that middlemen can own the distribution channels and siphon money of both developers and end users. The wayland protocol would form part of that as a way of isolating "Apps" from each other rather than providing the well integrated computing platform that was/is one of the drivers behind X11.
@Fierelier: yes, the Devuan repository is an "on-the-fly merge" of Debian packages that uses the packages directly from Debian repositories (except for the "forked" packages). I.e., Devuan only maintains and builds forked packages, and relies on Debian doing that for the rest.
In essence Devuan is simply a re-indexing of which packages are included and where to download them; forked packages from Devuan and all other packages from Debian. Thus, if Debian stops maintaining and building i386 packages, Devuan will follow.
bootlogd, if installed, kicks in after the devnodes are set up.
Iif you have another computer, you could set up a netconsole and then capture it there. Possibly with some clever hands-on, you could point that to a localhost console started (as early as possible) by initrd scripting. That would of course only capture from when that console starts.
Any bootloader (or earlier) output before the kernel is started might require a video camera, possibly even a fast-recording one.
Yes, that's due to the infamous "usrmerge" nonsense of Debian... they are in the process of moving stuff from /bin into /usr/bin because they think that everyone has set up links like /bin -> /usr/bin... and more. Web searching will tell.
You could have used --merged-usr for debootstrap and it would have configured the root filesystem with those stupid-links.
To come to there afterwards involves some manual copying hands-on and in particular a serious quirk in order to change /lib/ to a link without losing sight of the elf loader (e.g. /lib/x86_64-linux-gnu/ld-linux-x86-64.so.2).
It might be possible to debootstrap if you add
--exclude=logrotate,cron,cron-common-daemonto the command.
Ok. I'm really not sure which efivars files (aka "variables") can be deleted so a to recover space, but at least the most important ones are "immutable" (see with lsattr). Perhaps there are a couple of unused boot options?
You should back them up before experimenting!
Ah yes, you also need to
mount -t efivarfs none /sys/firmware/efi/efivarsYou'll need to mount the EFI partition on /boot/efi.
Note that the /dev/dri/* devnodes should be in group video and the same for /dev/fb0; group video with read+write access. They are made so by eudev, so the first guess would be that you are using some other hotplug handler which fails to do so.
The second part of that is that the user also needs to be in group video.
"the mail system at host april.topoi.pooq.com" has broken DNS.
cryptsetup-initramfs is not a forked package, so you should report to bugs.debian.org rather.
Perhaps if you avoid removing libpulse-mainloop-glib0 it might not remove xfce4 and hopefully retain that which you will want to retain.
btw, such output you should rather wrap as "code" than "quote" which would have made a scrolling element instead of a full two pages with such "nothingness".