[lsb-discuss] Re: REVIEW: System Initialization

Tobias Burnus tobias.burnus at physik.fu-berlin.de
Fri Jun 13 06:04:17 PDT 2003

Hallo Thorsten, hi all,

> > http://www.linuxbase.org/spec/wip/wip/sysinit.html
> Is there somewhere a version where the changes are marked or which
> could be easily diffed, so that the changes are visible?

cvs -d ':pserver:anonymous at cvs.sourceforge.net:/cvsroot/lsb' co -rLSB_1_3 \
    -d sysinit-lsb-1.3 spec/gLSB/sysinit
cvs -d ':pserver:anonymous at cvs.sourceforge.net:/cvsroot/lsb' co \
    -d sysinit-wip spec/wip/sysinit

and then a diff ...

But the main changes are:

* try-restart: restart the service if the service is already running
  [which is usually present in SuSE/UL init scripts]
  the try-restart option is optional

* [-p pidfile] is supported by pidofproc, killproc, start_daemon
  This makes it possible to start more than one daemon of a kind
  (e.g. one hightly restricted sshd with (only) root login and
   a normal sshd, two database server, etc.)

* Nailed down the exit codes, especially make sure that one can use
  the pidofproc, killproc, start_daemon in an init script without
  the need to check oneself whether the program is still running etc.
  [SuSE's killproc e.g. returns a "7" is the program is not running,
  in a case where the init script should return a "0" (SuSE bug 24909)]
     The idea of the exit codes is that start_daemon/killproc should
  use the exit codes which a init script has to return for start/stop
  and pidofproc the exit code of status. At least "0" = everything is ok
  and not 0 otherwise should be supported.

* killproc accepts also the pathname
  this allows killproc to use /proc/$pid/ checks which are not encouraged
  by the gLSB but de facto used by at least SuSE/UL's and Debian's

* Should-Start/Should-Stop are allowed, but a init script must not rely
  on its presents.
  [inspired by the footnotes from gLSB 1.[1-3] and

With best regards,


More information about the lsb-discuss mailing list