[systemd-devel] Antw: [EXT] Re: Still confused with socket activation

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Tue Feb 9 07:51:09 UTC 2021


>>> Andrei Borzenkov <arvidjaar at gmail.com> schrieb am 08.02.2021 um 19:43 in
Nachricht <63608ba2-0510-0e03-eb20-c92d86521695 at gmail.com>:
> 08.02.2021 12:10, Ulrich Windl пишет:
>> It seems systemd messes with that in a bad way.
>> 
> 
> Streetlight effect ...
> 
> For the last time - systemd does exactly what unit definitions tell it
> to do. Unit definitions belong to your application. If unit definitions
> that come with application are not suitable for your purpose, it is
> between you and your application.
> 
>> I suspect "Drop_ins"
> 
> As if systemd installs these dropins on its own.

Andrei,

I know you don't share my opinion on systemd, but anyway: Systemd is highly
complex, and maybe people (not talking about me; talking about distribution
makers) did not fully understand that complexity, shipping (well let's call
them) "non-perfect units" combined with the "trend of time", namely poor or no
documentation. The end user has to do a lot of guessing work, what the Units
are expected to do, what they really do, and which units are expected to be
enabled in what cases, what the dependencies are and what the correct order of
starting (in the manual case) is.
Maybe in an ideal world every thing would just happen "automagically" ("do the
right thing"), but we are far from that.
And understanding the complexity of systemd is much harder than understanding
classic init.
As for the drop-ins: I neither know what those are expected to do, not who
adds them at run time. See "documentation"...

Regards,
Ulrich






More information about the systemd-devel mailing list