You are not logged in.
My point was about operational overlap, not architectural equivalence.
In the specific case of linear aggregation, LVM is often used in practice instead of MD linear because it achieves the same end result (concatenation) while offering additional flexibility
“most admins” was shorthand for a general trend, not a statistical claim lol sorry about the imprecision
Don’t source /etc/X11/Xsession in .xinitrc and don’t use exec in .xsessionrc for background services
Should fix it
keep it minimal
Most admins today prefer LVM as it is more flexible
kernel maintainers are optimizing for less surface area and most common use cases. I would say MD linear fails in both
Why don't you build just the module, though?
What's your goal with that? It is designed to run as a daemon (or service) it'll probably assume devuan's layout
There is not a debian package for that. I guess you can clone it from the repos but its ankward
Run it as infra tool (container/venv) unless you really need a .deb
Yea, I tried it but it did not work. Tested many configurations but it couldn't even survive glmark2 without crashing my session. Besides Firefox crashed every 5 min even when disabling hardware acceleration. I'll just wait until the maintainers solve the bug I guess. I don't really need a compositor either way ;P
Small update: disabling the XFCE compositor seems to avoid the problem entirely here lol
At least on this machine, the regression looks related to compositing path rather than general display output.
The security model is interesting, especially the idea of returning dummy data
makes me wonder though, how do you plan to keep that predictable enough for client developers and debugging?
Posting this here for visibility. I already reported it against src linux and tagged it upstream.
After a kernel update on Devuan Excalibur, nouveau started misbehaving on my system with an NVIDIA GK208B / GeForce GT 730 [10de:1287]. Before that, the same machine was usable with nouveau.
What I am seeing is screen artifacts/pixelation, session instability, occasional return to login, and repeated nouveau GR/TRAP errors. The messages involve both Xorg and xfwm4, with repeated MULTIPLE_WARP_ERRORS and ILLEGAL_INSTR_ENCODING.
Hardware info:
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK208B [GeForce GT 730] [10de:1287] (rev a1)
Subsystem: Gigabyte Technology Co., Ltd Device [1458:3693]
Kernel driver in use: nouveau
Kernel modules: nouveauTypical log pattern:
nouveau ... gr: TRAP
nouveau ... [MULTIPLE_WARP_ERRORS]
nouveau ... [ILLEGAL_INSTR_ENCODING]One example from dmesg:
[ 588.244769] nouveau 0000:01:00.0: gr: TRAP ch 4 [007fa3c000 xfwm4[2583]]
[ 588.244777] nouveau 0000:01:00.0: gr: GPC0/TPC0/MP trap: global 00000004 [MULTIPLE_WARP_ERRORS] warp 0009 [ILLEGAL_INSTR_ENCODING]Right now I am testing with the XFCE compositor disabled to see whether that makes the session a bit less fragile.
If anyone else is on Excalibur with a Kepler-era NVIDIA card and nouveau, I’d be interested to hear whether recent kernels are behaving the same way for you.
Amazing release. Thank you, Devuan team.
Yeah in SDDM 0.21 the `--novtswitch` / `-keeptty` flags come from the logind code path.
They’re not in a config file, they’re baked in.
Devuan’s older SDDM (0.19 + elogind) doesn’t add them, so it still hops to VT7 i suppose
If you want Trixie to behave like that, either use elogind or patch/rebuild SDDM...
Not really the place for this kind of question but
Qalculate’s log() = log base 10
ln() = natural logarithm (base e)
You can redefine log() to behave as natural log via a custom function if you want, just:
alias log = ln(x)
Hey, your english is just fine don't worry aha
1. It's pretty literal. Devuan tracks Debian closely, same packages, same structure, same tooling but with systemd ripped out. Instead of systemd, you get sysv by default, but openrc and runit are also fully supported. Now, about libvirt: it works. You might need to tweak service files depending on how you launch libvirtd, but nothing show-stopping. incus, though, i would guess it will be trickiersince it’s a LXD fork and the LXD ecosystem has kinda bought into systemd socket activation and blah blah so you could maybe make it work with some shims or wrappers, but you’re really swimming upstream. 90–95% of packages work out of the box. Stuff that's glued to systemd will either:
not be in the repo
or need hacks
2. Yep. Devuan uses the same repo layout (dists/, pool/, Packages.gz, etc.) so you caan use rsync, apt-mirror, reprepro, whatever you like.
3. I cant speak for the devs but here's my thoughts: Debian could support multiple inits, but unless there’s a huge shift in developer interest, it just won’t happen. It’s systemd or bust, unless you jump ship lol (which you’re already doing). Upstream projects started assuming systemd APIs too, it seems (GNOME).
Note: i can't recommend enough debian's personal security guide, so consider following it when permanently migrating to devuan: https://wiki.debian.org/SetupGuides/Sec … alComputer
@zapper
Fully agree. Librewolf seems to be just firefox with some tweaked settings and uBlock origin tho.
Can't recommend Palemoon enough. It sucks a bit less than everything else, but it really depends on what you're looking for in a web browser
looool
if you're looking for something tailored specifically for Docker, something like s6-overlay or supervisord could be worth it for their simplicity and focus on process supervision in containers, from what I understand about them
Tried openrc with Void, but I'm a big sysv fan ;D
Not much to add either.
The challenge, as EDX-0 said, would be ensuring that these metapackages adhere to a certain level of standardization (e.g., following XDG Desktop standards). Otherwise, there's a risk that, for example, an Openbox-based "Bunsen Desktop" might clash with other desktop environments or not play nicely with certain XDG apps. But if these guidelines and standards are in place, it could offer a robust and flexible alternative to larger desktop environments like XFCE, GNOME, or KDE, while still maintaining a level of coherence and cross-compatibility. Assuming all this is desirable by the user.
A framework could be created. It should aim to provide a lightweight, modular desktop environment that adheres strictly to XDG Desktop standards for maximum compatibility and flexibility. The core package would install essential components like a minimal window manager, app launcher, and file manager, all adhering to XDG Base Directory and autostart specifications. From there, users could install specific modules for different window managers (e.g., Openbox, i3, Fluxbox), panels (XDE-Panel), docks (XDE-Dock), and XDG-compliant applications. This modular approach would ensure that components can be swapped without breaking system coherence. The framework would support seamless integration of both lightweight and full-featured applications, with automated testing and community-driven contributions ensuring ongoing compatibility. This structure allows users to build a highly customizable, performance-efficient desktop while maintaining cross-environment consistency and low overhead, providing a minimal yet robust alternative to heavier desktop environments like GNOME or KDE.
Look, I get the idea behind it, but honestly, it just feels like overcomplicating things. Linux is about flexibility, not pre-configured workflows. If you need a low-level programming environment, just install the tools you need and get on with it. Why add another layer of complexity with workflows? That’s just going to lead to more problems down the line when things break or packages get outdated.
The whole point of Linux is that you have the freedom to set things up how you want. You don’t need some workflow package telling you how to do it. If you really want to help people, just share your setup, share your dotfiles, maybe make a script—let others figure out how they want to configure things. Predefined workflows are a shortcut, but shortcuts rarely work in the long run.
So yeah, I’d say skip this workflow idea. Linux is about choice, not about boxing people into some “easy” solution.
Can't speak for the developers, but I would imagine that the reason sysvinit doesn’t auto-generate init scripts is simple: it's not as easy as you think. Every service has its own unique needs – paths, variables, dependencies, permissions. If sysvinit tried to auto-generate scripts, it’d end up missing some of that and make assumptions that just don’t fit all cases.
Imagine this: A service might need special user permissions, or it could rely on other services to start first. Some services need specific arguments. If sysvinit tried to do that automatically, it’d probably screw up.
That’s why you write your own init scripts. It gives you control, which is what Linux is all about. Everyone’s system is different, and you should be the one deciding how things work, not the system. The whole idea of Unix (and by extension, sysvinit) is to keep things simple, modular, and flexible. Each program or service is designed to do one thing well, and the system as a whole allows users to assemble these components however they see fit.
So, yeah, auto-generating scripts sounds nice in theory, but in practice, it’d be a pain in the ass. It’s better to leave that part to the user, I think.
X server might be having some issues, which is causing those crashes with xournal and sylpheed. From the log you shared, it seems like there might already be an X server running, which is blocking things from working properly.
First off, try checking if any X server processes are still running, and if so, stop them before trying to restart X. You can also check the Xorg log at /home/vb/.local/share/xorg/Xorg.0.log to see if it’s throwing any more specific errors.
If that doesn’t work, you could try reconfiguring X by reinstalling the xserver-xorg package—sometimes that resets things and clears up the issue. Also, it might be worth running apt-get install -f to fix any missing dependencies that could be causing problems.
Lastly, just to rule out anything with your user settings, create a new user and see if the issue still happens there. If it doesn’t, it’s probably something in your current user config causing the crashes.
Let me know how it goes!
Yeah, TP-Link's 10GB network cards, like the TX401 model, are generally compatible with Linux, including Debian-based distros like Devuan. The TX401 supports Linux kernel versions 3.10 and newer, so it should work with both Devuan 4 and Devuan 5, as long as you have the right drivers installed I think
Had the same problem, great to see it was easier than what I was expecting! I'm happy with OpenVPN and just using mullvad servers though. I don't trust those kinds of "custom" VPN software
Hey,
To be a sysadmin, you definitely need a good amount of knowledge and experience. That can come from self-teaching and real-world experience, not necessarily a degree. It’s not easy, though. That being said, don't waste more time thinking about an obstacle than it takes to overcome it
As for poker, maybe you're a pro, but it’s not always the most reliable way to make money. I’d say try blackjack, but hey, I'm a little biased! 😄 Learn the strategy, know the rules, and work the variance—it could make you some cash.
[Micronaut]
I'm not sure what tweaks are you (specifically) referring to, but I'll send some additional options I use on my Client.conf file, plus some justifications ;P
#My default "extra" options
auth-nocache # Stops saving authentication info, so it’s more secure
auth SHA256 # Uses SHA-256 for checking data integrity
cipher AES-256-GCM # Applies strong AES-256 encryption for protecting data
tls-cipher TLS-DHE-RSA-WITH-AES-256-GCM-SHA384 # Chooses a solid encryption method for secure connections
proto udp # Uses UDP, which is usually faster and can be more secure than TCP
persist-key # Keeps the encryption key around after restarts
persist-tun # Keeps the VPN tunnel up and running after restarts
dh dh2048.pem # Uses 2048-bit DH parameters for secure key exchange
# ecdh-curve prime256v1 # (Alternative) Uses Elliptic Curve for faster key exchange
user nobody # Runs the VPN service with minimal permissions
group nobody # Ensures the VPN service runs in a low-privilege groupIf using Proton with OpenVPN, I would check this guide on hardening: https://openvpn.net/community-resources … -security/
I might have some other configurations done, but that's what I remember now, hope it's useful
[GlennW]
Great insight! I tried it just now with no success though. It is possible that the VPN client is (somehow) already configured to use a different DNS server or maybe there are conflicts in DNS settings between the VPN and my system's network configuration, so the pushed DNS server might be ignored or something. My suspicion is that either resolv.conf or NetworkManager are overriding my VPN DNS settings.
I tried querying DNS records too and confirmed I was still not using the expected server, so the DNS leakage problem persisted.
And yeah, I'm staying away from Cloudfare. Still looks to me like a major data collection and profiling entity, despite their privacy promises. No relevant evidences though, just pure paranoia aha
[Micronaut]
Never looked into Proton since it's generally not highly regarded and has privacy issues... Not surprised I wasn't aware of that recommendation aha.
Yeah openresolv did it for me. Had some struggles in avoiding conflicts with NetworkManager, but after implementing the package and doing some fixes, DNS settings were (finally) getting updated properly when the network changed (i.e when I connected to the VPN). Which means, no DNS leakage! Thank you for the tip!
Still using DoT and DoH in the majority of my software, but now the VPN makes a more comprehensive protection against leaks at least.
Thank you guys :P