You are not logged in.
(You red it, don't you?)
Nope. Just a quick scan for now. When it's done, it belongs in the Documentation section.
Thanks for the write-up.
Before:
2:21.1.7-3+deb12u8devuan1
After:
2:21.1.7-3+deb12u9devuan1
apt policy xserver-common # same output for xserver-xorg-core
*** 2:21.1.7-3+deb12u9devuan1 500
500 http://deb.devuan.org/merged daedalus-security/main i386 Packages
100 /var/lib/dpkg/status
2:21.1.7-3+deb12u8devuan1 100
100 http://deb.devuan.org/merged daedalus-proposed-updates/main i386 Packages
2:21.1.7-3+deb12u7devuan1 500
500 http://deb.devuan.org/merged daedalus/main i386 Packages
I can only offer a data point - it's not happening here running daedalus on asus EEE with 686-pae kernel, minimal lxqt with no display manager. Tested before and after latest upgrade which gave me xserver-xorg-core and xserver-common.
Someone is working on fixing that issue. Until new isos appear, you can either install daedalus and upgrade to excalibur, or you can boot one of the daedalus live isos and do a debootstrap install of excalibur.
Or search this forum to read the other discussions about this for possible workarounds or if you want to help with testing/troubleshooting.
*.dpkg-dist is the saved copy of a new config file when the old one was kept in place. I think if you have brightness.dpkg-dist but not brightness, then something is wrong. Converseley, *.dpkg-old is the old version that's saved when it gets replace with the new one. I think you only get that when you tell debconf to use the package maintainer's (new) version rather than the old version that you modified from the default.
Sorry, I have no other answers.
More info -
I installed runit and made another iso. Autologin is configured by editing the run files in /etc/sv/getty-tty[1-6]. Autologin fails in this case, and the run files were not edited. In fact, it looks like live-config did not run at all - /var/log/live/config.log is empty.
This one boots to no login prompt, but pressing ENTER shows the login prompt and it works.
CORRECTION: The login prompt is there in runit, but I didn't see it because the console got spammed with system messages.
TTY autologin has stopped working in the excalibur live isos I'm making. This is normally done by the live-config script, 0160-sysvinit, which looks for a package named 'sysvinit' and if it's present, it modifies /etc/inittab to automatically log in the primary use on tty1. There are a couple of issues.
1. 'sysvinit' is a dummy transitional package. It's not a dependency of sysvinit-core, so it's possible to be using sysvinit as your init system without this dummy package installed. If it's not installed, then the live-config script doesn't setup autologin and the user is presented with a tty login prompt.
2. Now in excalibur, if the dummy package is installed, the system fails to complete its bootup. I can ssh in and see that /etc/inittab has been modified, but the ttys don't get created. I'm not sure where to look for the actual failure. The live-config script in excalibur is the same as the one in daedalus, where this problem does not occur.
I did a search on the debian-live mailing list and didn't find anything useful. What am I missing?
$ apt-file find init.d/brightness
initscripts: /etc/init.d/brightness
I don't think you can get rid of it. You're not the only one who never noticed this.
# Short-Description: Save and restore brightness level between restarts.
# Description: This script saves the brightness level between restarts.
# It is called from the boot, halt and reboot scripts.
I can't read most of that, but I know what it's tellilng you. Excalibur is still the Testing suite, so there is no -security or -updates repos. Those won't exist until excalibur is officially released as Stable. Comment them out for now.
To use one of the methods Andre linked, you need to get a list of just package names (filter out the versions) like this (or equivalet.):
SOMETHING BETTER THAN THIS!!! I didn't realize that 'apt list' just lists everything in the repo. You do not want to install more than 59,000 packages. See Delgado's post below or use 'dpkg -l' to see what's installed. (The awk would need to be modified.)
apt list | awk -F"/" '{ print $1 }' > package-list
I don't know if that's meant for making install lists. Feeding the list to apt install is possible, but might not be the best way. In the past, I've used 'dpkg --get-selections' 'dpkg --clear-selections' and 'dpkg --set-selections' to make an install list and use it on another machine.
I think 1161-openssh-server was a hack that I threw into Refracta isos and then later wanted to make sure it was gone. That's a good reminder of why I should use a hook script instead. Thanks.
The installer will recognize more than one existing efi partition and let you choose the one you want, so yes, you have to make it first.
In gparted, make a fat32 partition and give it the esp flag (it'll get the boot flag automatically).
Your motherboard might not let you be the one to decide which efi partition is the real one. And I don't know if having two and switching them by changing the drive order will work or is safe. The usual procedure is just to let linux use the same efi partition as windows and put itself in first place for booting. If you've looked at what's in the efi partition, you can see that there are several bootloaders. You can change the order or manage them otherwise with efibootmgr - depending on how much the motherboard manufacturer felt like following standards. see rodbooks.com for more info on uefi.
Nice to see you again and good luck with it. Post again if you run into problems.
BTW, I don't think you can call yourself a noob anymore. I checked my old forum logs. 2014!
Edit: fixed type (the year)
Of course, having someone who knows where to look is even better. Thanks, dzz.
I'm not sure what's up with ssh. In the excalibur isos I've made recently, live-config keeps turning off password authentication. I resorted to making a live-config script to look for 'SSH-ON' in the boot command.
1. I find myself fighting with grub a lot more in the last few year. It might not be you. If you have another linux installed, the easy solution would be to install Refracta without a bootloader, then let the other linux find it and add it to the existing boot menu.
You should be able to install to any partition when there are several on the disk and or several disks. The graphical installer will show you lists of partitions with checkboxes. The text installer requires you to enter the correct device name. That's probably something you're doing wrong or misinterpreting. No way to tell from here without more information.
2. Daedalus is the current stable version of Devuan. It's the same as Debian Bookworm (only better). Excalibur is the testing suite, same as Debian Trixie, and Ceres is Sid, always unstable.
3. Command names and file locations differ greatly between the rpm and deb worlds. Getting the scripts to work with both would be a major undertaking.
Newer kernel:
Here's an iso with a minimal graphical system, daedalus (stable) with backports kernel, and I think it has some firmware installed. You could install this and then add whatever you want.
https://get.refracta.org/files/experime … 4_1806.iso
Excalibur isos will be coming soon, but it's still in Testing so there will be frequent changes and possible breakage until it's stable. (maybe by end of year)
Sorry I missed this thread when it was new. Running refractasnapshot without syslinux would result in the script failing because it looks for syslinux/isolinux files to put into the iso. It might be possible to run it and then chroot into the filesystem copy to make changes and then run the xorriso command manually to pack it into an iso file.
I don't have any arm hardware to play with this. If anyone figures out how to do it, please let me know. If someone feels competent enough to try, I can help with pointing where to look in the script so they don't have to follow every strand of spaghetti.
Do you need to install firmware-amd-graphics? If so, you need to remove the stuff from amd. The script probably has its own removal option.
In daedalus you have to add 'non-free-firmware' to the sections in sources.list to get firmware that used to be in the non-free section. (example: firmware-amd-graphics)
You might need both of those sections, depending on what other hardware you have. (e.g. nvidia, some broadcom)
hello again, i386.
I replaced/updated the final debian kernel with linux-libre from FSFLA.
https://get.refracta.org/files/testing/ … 4_1613.iso
No kernel source or headers with this one. I selected the kernel that gets all the updates and changes (short term support) and that's probably fine while excalibur is in testing. If I continue with these builds, I'll probably change to the long term support kernel. Or you could do that yourself with 'apt install linux-libre-lts'.
edit: almost forgot this. it's also in the same dir as the iso along with a gpg sig file. sha256sum
d03a863bfdf91425b6b6d640450b7bd2b67c7c9c94c68898b57f918b7ea63fe7 refracta_13_nox_libre_i386-20250304_1613.iso
lxb, I'd like to try your amd64 build. I don't have time to build my own right now. I'll send you an email.
Yes, a new thread for the post-installation issues would be good. That makes it easier to find the useful information in the discussion later.
There's no problem with fdisk reading the image file. The problem occurs if I dd the image to a usb stick and try to boot that on amd64 hardware. The kernel sees the usb stick when it's plugged in, but it doesn't recognize a block device.
Uh-oh. I just plugged it in again and it did get a device name. But the system is complaining about GPT header and said to fix it with parted. I did that, and now fdisk can read the partitions on the usb. It doesn't boot on my hardware.
Is there a way to boot the image in qemu or virtualbox? I tried but failed - what is the correct cpu type to use? pine64.org mentions Cortex A53, Allwinner A64/R18 but I don't see those in the qemu list of supported devices.
So instead, I mounted the image file and chrooted into it, installed xserver-xephyr and jwm, and I can get a graphical session. I also tried installing xfce4, but I couldn't get that to run. I also couldn't get gnome to run, but I don't know the command to start it.
I wish I had a pine phone to try it for real (and for other reasons). The mobian site talks about dd'ing the image to usb and booting it on x86 hardware. Is this possible with the movuan image? I did put it on a usb stick, but when I plug it into the laptop, fdisk doesn't see it. Syslog shows usb device, but it doesn't get a block device name. Is there some package I can install that will recognize the partitions on the usb?
git.devuan.org is free. You could put the code there. I don't know if there are any space constraints that would prevent storing images there, and I don't think you can build there.
In addition to backing up your data, you probably should think about moving it to a system that has continued support for the kernel and maybe a system that isn't changing as rapidly as ceres.
Which kernel are you using? Did you manage to get the last one before they removed the 32-bit kernels from ceres? I believe that was 6.10.9. Do you have an alternate strategy for dealing with kernel patches? If so, those of us with old machines would like to know more.
I get the same error as nixer re: function check_chown. It seems like it's confused about where the logfile goes. I added set -x and tee'd the output. That didn't get the last of it, so I also made a screenshot. On re-read, I see that nixer was in ceres and I'm doing this in excalibur, but I don't think that should matter.
text output and screenshot
https://get.refracta.org/files/misc/alphautilz.out
https://get.refracta.org/files/misc/alp … output.png