[systemd-devel] restart vs. stop/start

[Lennart Poettering]

On Sun, 22.05.16 23:19, Christian Boltz (systemd-devel at cboltz.de) wrote:

> Hello,
> 
> Am Sonntag, 22. Mai 2016, 21:48:02 CEST schrieb Reindl Harald:
> > Am 22.05.2016 um 21:03 schrieb Christian Boltz:
> > > I'm already using ExecReload= to reload the profiles (works fine),
> > > and hope all users actually read the documentation and use reload
> > > (and avoid restart).
> > > 
> > > Please read the paragraph above the ^^^ marker again.
> > > The problem is what happens when someone accidently uses restart
> > 
> > when someone is using the wrong command - well, he is using the wrong
> > command - that's it - acting as root implies someone knows what he is
> > doing 
> 
> Did you ever hear of "better safe than sorry"?
> 
> *Every* admin will make an error sooner or later, so it's good style to 
> limit the damage as good as possible.
> 
> In case of AppArmor, the good old initscripts basically mapped "restart" 
> to "reload", so there was no risk of using the "wrong" command because 
> both worked in a sane way (= reloading the profiles).

This is a corner case we really don't support in systemd. Service
restart is always implemented by systemd, and we guarantee a fully
pristine execution environment for each service start. Individual
services may not override the restart logic. Sorry.

Lennart

-- 
Lennart Poettering, Red Hat