Systemd 254 released

Posted Jul 31, 2023 20:50 UTC (Mon) by sfeam (subscriber, #2841)
In reply to: Systemd 254 released by bluca
Parent article: Systemd 254 released

"ossification of known patterns" is a good thing. It is otherwise known as "stability". "broken" is another matter, but fixing something broken is not accomplished by dropping it altogether.

Systemd 254 released

Posted Jul 31, 2023 21:41 UTC (Mon) by bluca (subscriber, #118303) [Link] (2 responses)

Unit files are stable. Old and crufty init scripts were known for many things, but being stable is definitely not one of them. You will still be able to use them of course, if you like them - not on systemd though.

Systemd 254 released

Posted Aug 3, 2023 14:35 UTC (Thu) by jccleaver (guest, #127418) [Link] (1 responses)

>Unit files are stable. Old and crufty init scripts were known for many things, but being stable is definitely not one of them. You will still be able to use them of course, if you like them - not on systemd though.

Thank you for so clearly demonstrating my original point in this thread.

What you're calling "old and crufty" others call stability. Some of us (many of us) prefer imperative to declarative during important phases of operation because it makes it easier for a human to debug. Even leaving aside the fact that init scripts have been updated less and less due to systemd-migration, init scripts haven't changed much because they *just worked*... This is especially the case on the Red Hat/Fedora side of the ecosystem, instead of the Debian landscape where things actually were a bit more painful. But init script conventions in Fedora-world had been locked down pretty well for a long while.

Regardless, your "you can use them, but since I don't, you're not going to use them on systemd" attitude is *PRECISELY* what earns you such vitriol from admins downstream of your decisions.

I'm sorry YOU don't want initscripts on YOUR laptop, bluca. Some others of us would prefer that that compatibility layer stay. Thanks.

Systemd 254 released

Posted Aug 3, 2023 17:21 UTC (Thu) by anselm (subscriber, #2796) [Link]

I'm sorry YOU don't want initscripts on YOUR laptop, bluca. Some others of us would prefer that that compatibility layer stay. Thanks.

Nobody will prevent your init scripts from working. All that changes is that you will have to come up with your own systemd service units that call them. If your service is running now you can copy a simple one from /run/systemd/generator.late for later.

This is independent from the fact that on a systemd-based machine, when you're using init scripts – especially with a very simple (auto-generated) service unit – you're deliberately forgoing a bunch of cool and useful features which systemd provides out of the box, which could make your service easier to control, more stable, and more secure, and which SysV init files generally do not bother with. Of course this is your privilege, but it may be worth considering whether making a dedicated systemd service unit for your service which works without the init script and takes more direct advantage of systemd's features might not be a good idea after all.