On Mon, 22.08.11 19:54, Richard Shaw (hobbes1069@xxxxxxxxx) wrote: > > On Mon, Aug 22, 2011 at 3:53 PM, Lennart Poettering > <mzerqung@xxxxxxxxxxx> wrote: > [SNIP] > >> Is this all I need? It does not depend on any other service that I know of. > > > > Depends. If this daemon provides some interface to local clients you > > either want to make it socket activatable (requires patching) or use > > "Type=forking" or so, so that we can properly order bootup. > > Can you elaborate a little? > > What kind of patching are we talking about? Does the daemon need to be > systemd aware for socket activation? You don't need to patch, if you use Type=forking. > If I go with forking, am I ready to go? I don't know your daemon and its parameters. If you use Type=forking then the daemon should actually double fork. It should return in the parent only after the dameon initialization is finished. The daemon should write a PID file (and that PID file systemd should be informed about with PIDFIle=...) Type=simple just spawns off a daemon and does not wait for anything, continues immediately with the next service to strt. Type=forking spawns off a daemon and waits until it exits under the assumption that it forked off the main daemon in the background. Traditionally this is how Unix daemons notified the SysV init scripts that they were now fully started up. Also see daemon(7). Lennart -- Lennart Poettering - Red Hat, Inc. -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
