[systemd-devel] What if a service is started manually

[Cecil Westerhof]

On 11/29/2013 12:59 AM, Mantas Mikulėnas wrote:
>  >>  > In a trial presentation I used the following service file:
>  >>  > [Unit]
>  >>  > Description=Virtual Distributed Ethernet
>  >>  > After=syslog.target
>  >>  >
>  >>  > [Service]
>  >>  > Type=forking
>  >>  > PIDFile=/var/run/vde.pid
>  >>  > # Note the -f: don't fail if there is no PID file
>  >>  > ExecStartPre=/bin/rm -f /var/run/vde.pid
>  >>  > ExecStart=/usr/bin/vde_switch --tap tap0 --mode 0660 \
>  >>  >  --dirmode 0750 --group qemu \
>  >>  >  --daemon --pidfile /var/run/vde.pid
>  >>  > Restart=on-abort
>  >>  >
>  >>  > [Install]
>  >>  > WantedBy=multi-user.target
>  >>  >
>  >>  > Here the PID file is removed before the service is started.
>  >>  >
>  >>  > This brought up two questions.
>  >>  > - What happens is you start a service that you already started?
>  >> Nothing, or is the service first stopped and then again started?
>  >>
>  >> 'systemctl start' only starts services, therefore it will do nothing if
>  >> the service is already started.
>  >>
>  >> 'systemctl restart' would stop it and start it again.
>  >>
>  >>  > - What happens if someone started the service manually? So bypassing
>  >> systemd and running directly /usr/bin/vde_switch.
>  >>
>  >> As far as systemd is concerned, nothing happens - the manually started
>  >> vde_switch is just another process inside your login session. It will
>  >> *not* be automatically pulled into a "service" just because the program
>  >> name or something happens to match...
>  >
>  >
>  > I should learn to ask my questions better. T_T
>  >
>  > What I mend to ask. Someone starts /usr/bin/vde_switch manually and
> after that uses systemctl to start it.
>  >
>
> If the second vde_switch instance is configured to listen on the same
> sockets, etc., then... Well, it depends on the daemon itself:
>
> * most will consider this a fatal error, and exit with non-zero status,
> causing the systemd .service to fail as well;
>
> * but some will think that the existing socket is stale, will remove it,
> and happily start "on top of" the first instance. (Only happens with
> Unix sockets, of course; if the daemon uses TCP or tries to grab the
> same 'tap0' interface or such, then it can only fail.)
>
> I don't know how vde behaves. It will probably refuse to start.
>
> The best way to find out, of course, is to try systemd yourself.

I need only a generic answer. So if this question is asked I can answer 
with it depends on the service. Making sure that when they ask further I 
remember your explanation.

Thanks.