Pinning runlevels to conf.d folders

damobrisbane · n00b Joined: 20 Jan 2022 Posts: 9

Hi,

Does anyone have an opinion on the idea of pinning runlevels to conf.d folders. I find having /etc/conf.d as a fixed folder name as limiting.

ie given a runlevel tree such as:

/etc/runlevels/A/10
/etc/runlevels/A/20
/etc/runlevels/A/30

/etc/runlevels/B/10
/etc/runlevels/B/20
/etc/runlevels/B/30

then

selecting runlevel A >>> /etc/conf.d/A/...

or

selecting runlevel B >>> /etc/conf.d/B/...

or the like. cheers

figueroa · Posted: Fri Jan 21, 2022 4:49 am Post subject:

Incompatible with updates that subsequently write to /etc/conf.d/ making it high maintenance and a nuisance.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi

grknight · Retired Dev Joined: 20 Feb 2015 Posts: 1652

This feels like an answer in search of a problem.

I don't see how it can benefit anyone.

If the real problem was presented, then there is probably an easier way to do whatever is necessary.

damobrisbane · n00b Joined: 20 Jan 2022 Posts: 9

I am wanting to run up different configurations of the OS at different times. For example a basic profile or a monitoring profile. So I was thinking of a tree like runlevel configuration. A different softlevel is chosen at boot, at that will bring in a different set of configurations, like A/B configuration.

C5ace · Posted: Sat Jan 22, 2022 11:26 am Post subject:

I dual boot. Gentoo-1 for normal use. Gentoo-2 for experiments. Each uses different / 50GB partitions. The /home partition is the same. /var/cache/distfiles and/var/db/repos/gentoo/ are located on /home.

Disadvantage: This setup requires to run my update script twice.
_________________
Observation after 30 years working with computers:
All software has known and unknown bugs and vulnerabilities. Especially software written in complex, unstable and object oriented languages such as perl, python, C++, C#, Rust and the likes.

pa4wdh · l33t Joined: 16 Dec 2005 Posts: 811

I don't see a lot of use for something like this.

I have one service (cryptsetup) which i need twice with different configurations, and that was done by creating symlinks in /etc/init.d/ with different names and naming the conf.d files the same as the symlinks and adding the symlinks to the runlevels.
If one of the services you need doesn't work that way keep in mind that the files in /etc/conf.d are simply sourced by the init.d scripts so you can use all the bash features there. Nothing will stop you from reading /var/run/openrc/softlevel and set variables different depending on the value you found.
_________________
The gentoo way of bringing peace to the world:
USE="-war" emerge --newuse @world

My shared code repository: https://code.pa4wdh.nl.eu.org
Music, Free as in Freedom: https://www.jamendo.com

grknight · Retired Dev Joined: 20 Feb 2015 Posts: 1652

damobrisbane · n00b Joined: 20 Jan 2022 Posts: 9

Mistwolf · Posted: Sat Jan 22, 2022 7:51 pm Post subject:

figueroa · Posted: Sun Jan 23, 2022 5:20 am Post subject:

You should, perhaps, support your assertion that /etc/conf.d/ is overloaded. These are configurations used by the software, not all of which need user attention most of the time. If something in /etc/conf.d/ is unneeded, consider filing a bug for those packages, or volunteering assistance to developers/maintainers.

The option to do what you are suggesting does not exist unless you invent it. Dual boot, run an instance in a virtual machine, or set up a second/third/etc machine.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi

damobrisbane · n00b Joined: 20 Jan 2022 Posts: 9

Thank you so much for you replies. I didn't know that etc-update/dispatch-conf was even a command.

I'm going to keep plugging away at it, im happy with Gentoo/openrc. I guess one of my original motivations was around logging/monitoring with a fairly complex mix of applications. By that I mean docker/fluentd/opensearch/kafka, among others. Im often challenged by issues in the stack, for example some poor configuration (my own), is thrashing fluentd, but the interaction is with kafka. I then find myself hand-holding an app, for example updating a fluentd configuration file to switch to debug level logging and doing that similarly across applications where the problems are.

A couple of challenges in this scenario is manipulating the log level for a number of different applications at the same time. Unfortunately with some of these apps, they seem to consume quite a bit of resources, especially when running a number of them with debugging turned up and also some issue at play. Also, I am only wanting to debug for a limited amount of time. My goal really is to fix issues with as much information as I can, but once the issue is fixed, I want to go back to a lean as possible system as I can, even with logging completely turned off.

It was from this that I had a thought of having different softlevel targets, pretty much three: a failsafe level, an operational level and a debug level. I like the idea of a failsafe, a separate process is monitoring the (rc) system, and if something goes really bad, reboots on the failsafe runlevel. When I hit a problem, I run on a debug runlevel.

Cheers

figueroa · Posted: Mon Jan 24, 2022 5:36 pm Post subject:

It sounds like you may not be aware that to change an application's configuration, i.e. log level, verbosity, etc, never requires a reboot in Linux. Ordinarily just shut down the application and restart it. On rare instances if the application is integrated with the desktop environment, log out and log back in without rebooting. Reboot to use a different kernel. That's it.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi