Systemd 256.1 Fixes "systemd-tmpfiles" Unexpectedly Deleting Your /home Directory

[intelfx]

Remember the suspend-then-hibernate thing? This is what Luca Boccassi wrote then:


Personally, I have hit a few other (old or new) issues lately, some examples reported by others:

https://github.com/systemd/systemd/issues/11384

https://github.com/systemd/systemd/issues/28439

https://bugzilla.redhat.com/show_bug.cgi?id=1739689

Most of these have been fixed or can be worked around, but it's still irritating. I've migrated most of my systems back to OpenRC for now which at least is more predictable, I'll see how long that lasts.

So you've hit a few (pretty specific) issues in systemd-{network,resolve}d (which are absolutely incidental to systemd's init and very much optional), and you've decided to change the init to OpenRC in response to that?

Is it just me, or is there really no logic to be seen in this action?

[davidbepo]

Haters gona hate , thats how techonogy will be improved. If you dont like systemd just dont use it

thats the problem, you cant not use it

[klara]

So you've hit a few (pretty specific) issues in systemd-{network,resolve}d (which are absolutely incidental to systemd's init and very much optional), and you've decided to change the init to OpenRC in response to that?

Is it just me, or is there really no logic to be seen in this action?

Sure, none of these issues are anything major and I probably could have resolved them without switching from systemd. But for me the primary benefit of systemd is the ecosystem it brings, and if that's problematic and needs to be worked around, why bother? As you may have seen in my message, I even leave it open that I may switch back. But generally, OpenRC seems to work well enough and I don't particularly care about my init system as long as it does not cause me too many issues.

I've had another issue which is probably related to my custom initramfs which causes systemctl daemon-reexec to fail and freeze systemd which I haven't had the time or will to debug more which is also contributing to my decision but decided to not list that before as it's probably my fault. It does work with OpenRC though, including dm-crypt which I was never able to get to work properly on systemd.

So yes, there may not be a entirely logical reason for switching to OpenRC, but it wasn't particularly difficult either and I can always switch back if I want.

[macemoneta]

Why is no one actually addressing the real problem this issue highlights:
"Including the /home that is created by systemd-tmpfiles' home.conf. "

What kind of stupidity is this?
Why are a function or command to create temp-files used to create /home? WTF?

Is this done by a distribution or bad system administrators, or is this intended or recommended way from the systemd developers/projects?

If it's the first, a note on the spesific distro had been nice. To avoid something made by incompetents.

If it's the second, the systemd developers are even worse than I thought and they deserves much worse than he comments in this tread.
That is beyond stupid!

I just checked, and on Fedora (40) '/home' is created by the 'filesystem' package, and /etc/tmpfiles.d is empty. So this definitely seems like a distro-specific issue.

[User29]

Haters gona hate , thats how techonogy will be improved. If you dont like systemd just dont use it

Also, RTFM

[fotomar]

Every time I read anything about systemd, I'm so happy I decided long ago to not use it.

[Rallos Zek]

Systemd walks into a bar, shoots the owner, proclaims it is the new owner, turns the bar into a casino, adds a hotel, a grocery store and an init system.

1ddbha.jpg

[pWe00Iri3e7Z9lHOX2Qx]

I like systemd overall, but some of the devs are real jackasses. The tone and wording of that response reminds me of the systemd sleep.conf change where the dev basically told everyone to fuck off and they were using it wrong, even though the sleep then hibernate usage was an obvious and hugely useful use case for s2idle laptops.

[bosslog]

https://github.com/systemd/systemd/issues/25269

https://github.com/systemd/systemd/issues/33349

> systemd developer Luca Boccassi ("bluca") of Microsoft

The inane responses by "bluca" almost make it seem like Microsoft is making systemd bad on purpose to sabotage Linux...

[pWe00Iri3e7Z9lHOX2Qx]

https://github.com/systemd/systemd/issues/25269

https://github.com/systemd/systemd/issues/33349

> systemd developer Luca Boccassi ("bluca") of Microsoft

The inane responses by "bluca" almost make it seem like Microsoft is making systemd bad on purpose to sabotage Linux...

Ahh...so it's the same jackass from the sleep then hibernate thread. I guess it's good that there aren't two of these yahoos. But maybe this guy shouldn't have anything to do with triaging bugs Microsoft.