I'd suggest to re-install Devuan with separate partitions for /etc, /var, /usr and /home. Preferably, put /home on a separate disk if you can. This prevents issues like you described where one (rogue) process fills the entire disk and your system is unworkable.

Furthermore, I'd suggest to use the task-mate-desktop (or if you must, use task-lxdm-desktop) as a package after installing a basic system. This task will take care of all requirements to get to a working Mate (or LXDE) desktop.

Finally, I'd suggest to obtain a bootable copy of the SysrescueCD, which enables you to repair Linux systems that have been broken by user-(in)action.

Perhaps worth noting I never got the "Debian way" of building a kernel to work (i.e. creating a .deb kernel package), I always used the 'direct' method. It also allowed me to install the binary nVidia GPU drivers fairly easily, at the time ATI (later part of AMD) wasn't too keen on providing Linux drivers for their GPU range.

I guess you could call on M$HQ with a stack of re-writable DVD's and d/l the lot there. You tell them that Bill won't mind

(sorry, that was too easy )

Post the results of the following commands (root required!)

dpkg --get-selections > selections.txt
rc-update show > rc-level.txt
uname -a

You can copy the results of the last command directly, but the other commands each produce a text file. Post the contents of these files.

Way back before Devuan existed I'd build my own kernels. So much so, I created a script for it:

#! /bin/bash
#
# Small script for building a new 3.x kernel on Debian and derivatives
# Written by <private>
# Version: 1.0 @ May 6, 2014
# Requires minor editing for latest stable kernel version
# The make install and mkinitramfs steps requires root permissions.
# This script is released under the GPL: http://www.gnu.org/copyleft/gpl.html

# Preparing your system:

# First, we'll install the build environment with some extra's to aid configuration of the new kernel

apt-get install build-essential libncurses5-dev wget tar

# To keep things tidy a new directory for the kernel is created. Remove or comment out both lines if you don't want this.
# If you comment these line out or remove them, the kernel will be downloaded and installed in the current working directory.

mkdir ~/kernel
cd ~/kernel

# Getting the source and preparing it for use:

# edit these lines to represent the latest (longterm support) stable kernel from the kernel.org team

wget -q https://www.kernel.org/pub/linux/kernel/v3.x/linux-3.12.18.tar.xz
tar -xvf linux-3.12.18.tar.xz
cd linux-3.12.18

# Configuring the new kernel:

# Run manually if this part fails and accept all defaults, then comment out the next line with a # in front.
# Alternatively, use "make oldconfig" but this will ask a lot of questions you'd need to answer individually.
# Provided you have installed the correct packages, you can also use the "make xconfig" tool if you're working
# in a graphical environment.

make menuconfig

# Building the kernel and putting everything in place to use:

# This is the part that does the actual work:

make && make modules && make modules_install && make install && mkinitramfs -o /boot/initrd.img-3.12.18 3.12.18

# The last portion of the above line should be edited if another kernel version is used!

# Now you need to update your bootloader to add the new kernel. Most Linux distributions now use Grub2 by default.
# If yours uses the old Grub or Lilo, comment out this line and configure/run manually.

update-grub

# After you've done that, and checked thoroughly twice, it's time to reboot into your new kernel.

exit 0

As you've noticed, this is aimed at the 3.x series kernel, I guess with minor editing this script can also be used for new 4.x and perhaps even the upcoming 5.x series kernel. Unless the tool-chain or its syntax has significantly changed since, I haven't had to build a kernel in several years, due to using a (cutting-edge) source-based distro. (Funtoo, in my case) This script was also intended as a teaching tool for those wishing to compile their own kernels but hadn't tried yet, hence the copious amounts of comments.

HTH!

[edit: I've also found an earlier, undated version of this script, dealing with a 2.6.38 kernel (that's telling! ) last modified Spring 2011...   [/edit]

It's related to building the kernel module for the i915 GPU driver and it's a trace (as you've already noticed) for the developer(s) which shouldn't really be there. Do you have any related .dev package installed on your system? It may have 'escaped' from there. Report it as a bug to whomever is/are the dev's for the Intel i915 driver.

Here's the key error from your log:

[    6.201782] xc2028 4-0060: Incorrect readback of firmware version.

Ensure you have the correct firmware extracted from the script, perhaps download the latest Win-OS driver (again) to restart the extraction process. You also mentioned you didn't/don't have the directory /dev/dvb, try creating that manually and rerun the extraction script to see if it puts the various firmware blobs in the correct place.

If memory use is an issue for you, don't use Devuan or any other main distro. Instead, switch to a lean system like LFS (Linux from scratch). Having said that, I'd recommend updating your hardware, starting with the old harddrive. Refurbished drives don't cost a lot and even new ones don't break the bank anymore. BTW: Linux uses as much RAM as it can find and has pretty good memory-housekeeping in the kernel.

As for Grub: yes, it takes time, especially on older hardware (which you implicate you have).

I see no reason why it couldn't, given that Devuan, like Ubuntu, is based on Debian. The only hurdle might be if OpenStack depends on some specific systemd stuff that's not (yet) ported to Devuan, so best check the requirements of Open Stack and see if Devuan offers the functionality it needs.

Well, that's the problem: not everyone has fast broadband access to the web at the place (s)he wants to run Devuan. So, for my work I do the following: at a location with fast broadband.access (at home) I create/update a local mirror, then copy said mirror to an external HDD, take this HDD to my place of work, where I don't have access at all, then copy the mirror from the HDD to the local mirror on the server I set up for this purpose. I can assure you rsync is a godsend time-saver on copying files over   This setup allows me to install Mate offline on refurbished PC's, w/o resorting to switching disks all the time. I simply install a basic system, then modify the sources.list file to use the local offline mirror exclusively and install from there.

In that case, you'd need to provide a boot-script for OpenRC that contains the webmin start command. There's plenty of info on how to write a boot-script for Open RC on the web, or simply copy an existing script, rename and edit it properly then invoke the openrc command to start it.

When installing Webmin, it doesn't automagically configure itself to start at boot. Easiest way is reinstalling the latest Webmin (1.910 at the time of writing) then head for "System" -> "Bootup and Shutdown" in the menu, mark Webmin to start at boot and apply.

[edit]Alternatively, if you can login on the cli via ssh, run the command (as root!):

sh /etc/init.d/webmin start

This enables you to skip reinstalling Webmin, but you still need to instruct it to start at boot as described.[/edit]

HTH!