You are not logged in.
By commenting out that line, the start job still ran in the live session; however, its time was cut in half...about 2-3 minutes before booting into the desktop...as opposed to 5-7 minutes.
Was the error message the same for both images?
Check the output of my journalctl command for both.
I'm a bit busy at the moment but try a rebuild again later.
The package is back in Debian buster with a new error message for Wayland users and an added dependency on zenity:
https://packages.debian.org/source/buster/synaptic
I presume it will be added to beowulf soon.
What is your graphics hardware?
Exe GNU/Linux doesn't include any non-free firmware but AMD, NVIDIA and recent Intel graphics cards require that to work properly.
And did you verify the md5sums after downloading the image?
You can check for a bad USB stick by copying the image back to a file:
# cp /dev/sdX check.img ; sync
And then compare the checksums again.
A start job is running ...blah...blah...blah... no time limit
That appears to be caused by /lib/live/config/0040-sudo, specifically this line:
sudo -u "${LIVE_USERNAME}" sh -c "echo 'SU_TO_ROOT_SU=sudo' >> /home/${LIVE_USERNAME}/.su-to-rootrc"
I just tried unpacking your ISO, commenting the line then repacking the image and it then boots without error (in QEMU).
No idea why it causes the problem but the systemd journal helped me find it
For the record:
journalctl -u live-config
LOLOLOL!
Still waiting to hear of someone migrating a devuan install to debian.
https://dev1galaxy.org/viewtopic.php?pid=15566#p15566
Challenge accepted!
Devuan beowulf booting with systemd as PID1:
Unfortunately though I can't get any input devices to load for X so it's console only (the same situation occurs if I boot with sysvinit in Debian buster).
I had to install a version of eudev without the postrm & prerm install scripts and remove the eulogind package in a chroot (from Alpine Linux) and a few other corrective measures.
Don't worry folks, it was all done in a btrfs snapshot — I still have my original, fully functional beowulf system
I'm not sure if Devuan supports fingerprint recognition or not
They have fprintd in the repositories so it should do.
I would test it on my ThinkPad X201 but it pulls in polkit and there's no way I'm having that crap on my system
Can you help me to get my system in Legacy mode?
First create a BIOS boot partition on your main disk: I like to use gdisk and add a new partition of type "EF02" in sectors 34-2047 (these should be free in a correctly aligned disk), do not format the partition.
Then install the grub-pc-bin package and run
# grub-install --target=i386-pc /dev/sdX
Replace X with the letter assigned to your main drive.
Finally reboot and make sure that CSM or "Legacy" mode is enabled in your firmware ("BIOS") options, this will emulate a non-UEFI system and should allow memtester86+ to work.
How can I manage grubx64.efi entries under /boot/efi?
I tried efibootmgr but seems useless—some entries are rewritten at boot?
If your firmware is crap (most are) then you will probably be using the fallback UEFI loader located on the EFI system partition at /EFI/BOOT/BOOTX64.EFI
See https://www.rodsbooks.com/efi-bootloade … iples.html & https://www.rodsbooks.com/efi-bootloade … ive-naming for more on this.
Added a statically-linked version (useful for rescue situations in which /usr is not available) and also included all the hardening flags (-fstackprotector-strong etc), version bumped to 6.4-2
mkdir -p ~/builds/dcaenc
cd ~/builds/dcaenc
wget http://aepatrakov.narod.ru/olderfiles/1/dcaenc-2.tar.gz
tar xf dcaenc-2.tar.gz && cd dcaenc-2
./configure --prefix=/usr/local
make
sudo make install
It was pretty simple to package it up for Debian/Devuan so here it is:
https://software.opensuse.org//download … age=dcaenc
Disclaimer: untested so let me know if it works
The open source version doesn't work in UEFI mode.
It is possible to get a UEFI system booting in non-UEFI ("Legacy") mode, post back if you want help with that.
For a UEFI-compatible version see https://www.memtest86.com/download.htm
Otherwise, try memtester instead, that can be run from the booted system.
It was possible to switch from sysvinit to systemd as PID1 in wheezy but sysvinit was the default.
You could use a Debian buster "live" ISO image to install Devuan ceres.
Load up the desktop then make sure debootstrap is removed:
# apt purge debootstrap
Then download the Devuan debootstrap package from here:
http://deb.devuan.org/devuan/pool/main/ … n1_all.deb
And install it with
# dpkg -i debootstrap_1.0.114+devuan1_all.deb
You can then follow this guide to install the Devuan system (skip the "install debootstrap" section, obviously):
https://www.debian.org/releases/stretch … 03.html.en
For the debootstrap command use this version:
# debootstrap ceres /mnt/debinst http://deb.devuan.org/devuan
I used this method to install Devuan ascii from a Debian buster system, it should also work for ceres.
EDIT: a few shortcuts:
Rather than using MAKEDEV and mounting /proc you can instead install the arch-install-scripts package and run
# arch-chroot /mnt/debinst
to enter the chroot, this will bind mount all of the API filesystems automagically.
It also includes the genfstab command, which can be run before entering the chroot, like this:
# genfstab -U /mnt/debinst >> /mnt/debinst/etc/fstab
And instead of editing /etc/adjtime directly use this command:
# hwclock --systohc
Quilt rocks!
Patched the Makefile to rename the binary to /bin/loksh and added loksh to the ksh alternative system.
The package no longer conflicts with ksh93 and can be safely co-installed
New version is 6.4-1.1:
https://software.opensuse.org//download … kage=loksh
To set it as the default shell use either
chsh -s /bin/loksh
Or use /bin/ksh instead and also run
update-alternatives --set ksh /bin/loksh
to set loksh as the /bin/ksh alternative.
^ Thank you very much cynwulf, that is excellent advice.
And I've just realised that my package breaks the ksh alternative as well because it overwrites it...
I will try to patch the Makefile, I think that's the way forward here.
Everything seems so much more solid without systemd <hehehe>! The default desktop environment seems much better integrated.
I'm curious: does the "Time and Date" configuration work?
Menu → System → Time and Date
Reference: https://dev1galaxy.org/viewtopic.php?id=2770
Although bash is a fine shell with many useful features, it does have it's problems:
BUGS
It's too big and too slow.
OpenBSD uses their own KornShell implementation (derived from the old public domain KornShell, pdksh) as the default shell, the codebase is much smaller than that of bash and benefits from the audits performed regularly by the developers. It uses less memory than bash and is also faster[1] but doesn't support here strings or numerical brace expansion; here documents & standard brace expansion work fine though.
There are two Linux ports of this program available:
https://github.com/dimkr/loksh
They compile under Devuan so I have packaged them up and added it to my OBS repository (the oksh package was ported by an OpenBSD developer and so should probably be favoured):
https://software.opensuse.org//download … ckage=oksh
https://software.opensuse.org//download … kage=loksh
The repositories are technically for Debian but they should be compatible with Devuan and they only hold the loksh packages, there are absolutely no systemd dependencies at all.
The package currently conflicts with ksh93 because the binaries share the same name but I am researching how to change the name of the installed binary so I can remove the conflict (any hints welcomed), I will update the package when I figure it out. Fixed!
To change the default interactive shell run
chsh -s /bin/oksh # or /bin/loksh
The configuration file is set by $ENV so add this to ~/.profile and/or ~/.xsessionrc to make it read ~/.kshrc every time a shell is opened:
export ENV="${HOME}/.kshrc"
Note that the command line editing mode is determined by $VISUAL (or $EDITOR if $VISUAL is unset) so if vim is preferred then add this line to ~/.kshrc to restore the usual behaviour (ie, up arrow to recall the history, etc):
set -o emacs
[1] Output of ps_mem showing memory usage:
580.0 KiB + 96.5 KiB = 676.5 KiB ksh
2.4 MiB + 216.5 KiB = 2.6 MiB bash
Size of binaries:
-rwxr-xr-x 1 root root 1.2M Jan 24 10:01 /bin/bash
-rwxr-xr-x 1 root root 264K Apr 9 19:13 /bin/ksh
Rough benchmark for loksh:
$ time for i in $(seq 1 1000000);do [ 1 = 1 ];done
0m01.96s real 0m01.90s user 0m00.09s system
Bash:
$ time for i in $(seq 1 1000000);do [ 1 = 1 ];done
real 0m3.633s
user 0m3.586s
sys 0m0.061s
Bash with numerical brace expansion:
$ time for i in {1..1000000};do [ 1 = 1 ];done
real 0m3.944s
user 0m3.876s
sys 0m0.064s
EDIT: added oksh port.
^ Oops! Thanks for the correction.
@OP: sorry for the misinformation.
the adapter is called enp0s3 instead of eth0
See https://www.debian.org/releases/stretch … face-names & https://www.freedesktop.org/wiki/Softwa … faceNames/
tl;dr: either substitute the new name in /etc/network/interfaces or add net.ifnames=0 to GRUB_CMDLINE_LINUX in /etc/default/grub & run update-grub (as root).
devuan_ascii_2.0.0_amd64_netinst.iso
^ I think you should use this image rather than the i386 version, for the reasons I have already listed.
My machine starts for a second, shuts down again and starts a second time, after that procedure ist runs stable
Sounds like that may be a power supply problem, is there anything in the logs?
You can test the RAM with memtester.
Some ISO images have memtest86+ as a boot option (if started in non-UEFI mode).
The Debian live ISO sometimes causes problems when it is used to install the system, I always recommend the netinstall image.
Looks like perhaps the Devuan installer has inherited this.
GPT includes MBR boot sector in 1st sector, thats how the iso can boot in both old MBR (supposedly, in regards to OP problems) and new UEFI.
Although a GUID partition table includes a "fake" MBR it is not large enough to hold GRUB's core.img (which is stored in the MBR gap in msdos partition tables) and so a BIOS boot partition is needed.
ISO images use the El-Torrito "hybrid" configuration to accommodate both UEFI and non-UEFI booting:
https://wiki.osdev.org/El-Torito#Hybrid … _USB_stick
how you made bootable USB sticks? Did you use dd, unetbootin or another tool?
This thread is about a full system installation to a USB stick rather than ISO images.
So it seems problem is in USB 3.0 stick.
Yes, there are a few similar problems on the interweb:
https://superuser.com/questions/1268430 … her-system
https://askubuntu.com/questions/868744/ … om-usb-3-0
The second thread claims that a firmware ("BIOS") update fixed the problem, is that a possibility for your non-working machine?
assuming the USB is using GPT
A GPT disk (or stick in this case) would require a BIOS boot partition to start in non-UEFI mode and I don't think the OP added that.
old GPT entries were still on the USB
For reference:
# apt install gdisk
# sgdisk --zap-all /dev/sdX
^ That will remove both the main and backup GUID partition tables from a drive (replace X with the relevant drive letter ofc).
There is message that no boot media found
Is there an option to select the boot device?
On my ThinkPad X201 I get a boot menu if I press F12 during startup.
gfxmode lasted 15 seconds the first time but reverted back to one second.
That is very strange.
What if you set GRUB_TIMEOUT to "-1", is that honoured?