The Best #Linux Distributions Without systemd
https://www.howtogeek.com/713847/the-best-linux-distributions-without-systemd/
@rgegriff @lupyuen
my first Mastodon post 🙂 ...
Systemd can introduce problems, because it tries to integrate a lot of different services that were distinct and customizable in the past. I think to logginng for example: with Systemd you have integrated logging. But if you have a very big server, generating a lot of logs, it is difficult to customize the Systemd default logging system, while before Systemd you had a lot of different options from which choose.
Before Systemd, the service specification was less elegant, but Linux administrators were in full control, because they can swap parts. With Systemd they had to accept the choices of Systemd, and in some usage scenario, when they need to swap parts, it is not configurable/customizable enough.
It is a complete change of philosophy for Unx, because in Unix usually you assume to being in control of the details of the system, when you need this.
@lupyuen thanks! I'm studying different fediverse/P2P tools. Now I have doubts... if I hello to you, I'm using Mastodon as a chat, and is this good? 🙂
As Twitter alternative it seems rather good.
@mzan haha yep Mastodon is great! 🙂
@popolon @lupyuen yes, probably you know better than me. Thanks and sorry for the FUD.
I'm rather sure that in the past there were problems with Systemd and syslog integration, as reported here https://news.ycombinator.com/item?id=9359167 but they are mainly bugs, probably nowdays resolved, and not defect by design, so you are right.
A possible "defect by design" is the fact that Systemd in any case want to be the first logging service in the pipeline, and then it can redirect to other syslog daemons. But obviously if the code is not a disaster, it should not introduce new problems, because you can disable journalctl heavy operations, and then Systemd will do only a very simple redirect.
On Hacker News there were regular posts against Systemd, written by administrators. It seems that it tried to do too much, and too fast, introducing some incompatibilities, and not being rock-solid in its implementation. So it generated some hate, because it solved problems already well managed by experienced administrators, but introducing new bugs and integration problems.
Probably nowdays the benefits are more than the left problems. I don't know in details because I administer only few servers, and I'm mainly a programmer.
In any case, indipendently from Systemd, the task to init a system and services is a very complex one, and a tool cannot cover all possible usage scenario. In extreme cases I think that there will be devop tools, with a minimal Systemd automation.
@mzan Welcome to Mastodon! 🙂