[skissane]

I tend to downvote jokes or non-substantive comments about systemd (although I haven't downvoted any of the comments on this post).

Like any other software system, anyone is welcome to question or doubt or disagree with systemd's design decisions. But some people seem to have an almost irrational hatred for it, far beyond what a mere software package calls for. Even if a person making a joke about systemd isn't themselves like that, being reminded of it all makes me want to reach for that downvote button. And the topic gets tiring after a while anyway, it has been done to death already.

[denton-scratch]

> But some people seem to have an almost irrational hatred for it, far beyond what a mere software package calls for.

That could be because somehow it's managed to become the default init on all mainstream distros, leaving few good options for those who would prefer to avoid it. I don't want to use a distro whose main raison-d'ètre is avoiding systemd.

[edoceo]

Gentoo and others exist for far more reasons than just avoiding systemd.

[denton-scratch]

Hmm, I thought Gentoo defaulted to systemd. Rather, I learn that it's an 'alternative'.

https://wiki.gentoo.org/wiki/Gentoo_Without_systemd

Looks like you have to do a bit of tinkering to ensure it stays free of systemd. Same with Debian, which I use with SysV init. But it took a couple of hours tweaking to switch - perhaps one of the other inits might have been easier to switch to.

[edoceo]

My Gentoos have been OpenRC since forever. And maintaining that requires just one USE flag.

I've never tried to move debian to not-systemd.

[miohtama]

Systemd of course has had its share of issues. But a lot of also comes down as natural resistance of change by sysadmins, which is a well documented psychological phenomenom

https://journals.sagepub.com/doi/full/10.2466/09.02.10.CP.1....

… and sysadmins tend to be at the deep end of the spectrum

[bayindirh]

> But a lot of also comes down as natural resistance of change by sysadmins...

I find this "meh, you're just being stubborn" labeling is a bit dismissive. Especially when a lengthy technical comment about systemd is replied with the tone of "You just used to old ways eh? Time to freshen your knowledge, hehe".

Today's sysadmin dynamics doesn't allow one to stagnate with old knowledge either. Yes it's very useful, but not eternally. Instead it allows you to build knowledge faster.

[da_chicken]

It's not remotely unique to sysadmins. If anything, devs are worse.

Have you seen how many Python 2 packages there still are out there? How often do we see opinions on HN that amount to "C > *" whenever Rust, Go, Java, C++, etc. are in the submission? That's the exact same "resistance to change".

Change isn't a light switch. It doesn't go from 100% old to 100% new. Change is a long, slow, time consuming, and difficult process that requires extra effort at every stage in the process. There are several different stages:

1. Not knowing the new and being confused.

2. Identifying the new and having to develop special processes.

3. Accepting that the new is here to stay and translating or adapting the processes and procedures from the old, and spending effort to re-learn the same features in the new system to maintain business as usual.

4. Discovering all the new ways that the old process doesn't fit the new system or the new system isn't designed for your existing business processes, and fixing them one catastrophe at a time.

5. Running the old and new, simultaneously, for the lifespan of the systems that depend on them.

6. Having the new significantly overtake the old such that it's now unusual, and being frustrated that the old still exists.

7. Forgetting knowledge of the old via atrophy.

8. Encountering the old and being confused.

And that's when the new system proves itself to be better than the old.

[KptMarchewa]

>Have you seen how many Python 2 packages there still are out there?

Pretty much none. Can't say exactly how many, since http://py3readiness.org/ is dead.

[pid-1]

I'm a young-ish (late 20s) sysadmin and I've been reflecting on how I'm becoming increasingly cynical and resistant to change over time. I will probably organize my thoughts at some point, but here are some unsorted ones:

* Infrastructure is often untestable. I've been bitten by the silliest changes in unexpected ways. We use IaC extensively, which makes things more organized and easier to operate, but still, testing stuff is insanely hard and time consuming, often impossible.

* Infrastructure is often thightly coupled. Many architectural decisions will come back to haunt you in the future, in very serious and unexpected ways.

* Infrastructure is often shared. Consequently, balancing requirements from different teams with different goals is very hard.

* Often, I have no skin in the game. I will not earn more money to implement changes that will benefit my peers, but my life will suck more if I do. Often, being a intransigent bureaucrat is the most rational path I can take.

* Often, devs have no skin in the game. They will earn more money if features get delivered, but won't have to deal with the long term operational and security consequences of the changes they ask for.

That was already repeated ad nauseam by others, but my best professional experiences were:

* The times I was responsible for software and infrastructure of a project.

* The times I was working in a small team, very close to devs and thinking about a single project.

[cratermoon]

My dad was a systems programmer and he really didn't like change. His mantra was, "If it ain't broke, don't fix it".

[grayhatter]

I systemd because I deeply disagree with the philosophy of it's developers. You shouldn't write system critical software if you don't understand how and why said system works. I'll give you It's much better now, but I don't forgive easily. I till haven't touched debian after they disabled crypot-randomness either. I've also gotta say, it's a little annoying to see someone condense all the issues surounding systemd as "hur dur, change is scary... grok no like change"

[gmfawcett]

I agree, but it seems just as shallow to condense systemd to "it's much better now, but I don't forgive easily."

[rovr138]

I was with you at the beginning.

This part just takes away from your argument,

>… and sysadmins tend to be at the deep end of the spectrum

Source?

[tapas73]

> I was ...

We'll need a peer reviewed article to believe that.