Comment # 22 on bug 906900 from Thomas Blume

(In reply to Reinhard Max from comment #21)
> (In reply to Thomas Blume from comment #20)
> Or to put it all into a more compact form:
> As long as it claims compatibility with init scripts, systemd has no
> business in killing processes that were started by an init script unless it
> is explicity told to do so.

Ok, I didn't see any reasoning on the upstream list why the default KillMode
for sysv scripts is set to process.
I'd agree that normally the init script should control the process termination.
I'm going to change the KillMode to none.
We might buy system hang problems at shutdown, in case the init script hangs,
but this would be the same behaviour like on a sysv system.

This will also circumvent the bug from comment #20, so I don't need to address
this separately.
I feel confident to do so, because the only solution given is to revert the
commit, which deviates more from upstream than the KillMode change does.