[systemd-devel] I wonder… why systemd provokes this amount of polarity and resistance

[Zbigniew Jędrzejewski-Szmek]

On Tue, Oct 28, 2014 at 05:16:38PM +0100, Jan Alexander Steffens wrote:
> On Oct 28, 2014 5:05 PM, "Lennart Poettering" <lennart at poettering.net>
> wrote:
> > On Tue, 28.10.14 11:28, Dale R. Worley (worley at alum.mit.edu) wrote:
> > > That is not entirely true.  I'm a user (because systemd is in Fedora
> > > 19), and I've complained that if I mark an /etc/fstab entry as
> > > "nofail", some part of systemd will wait *forever* to see if the
> > > partition becomes available, whereas the behavior that I want (which
> > > was provided in earlier Fedora releases) is that once the system
> > > gets to the point of user logins, it will give up on automatic booting
> > > (and leave it to manual control).
> >
> > I have already replied to this, and pointed out that such a scheme is
> > inherently racy, and that this is something we will unlikely support
> > natively in systemd. Sorry for that.
> 
> I think the actual issue here is the behavior of Type=idle, which delays
> the gettys for an annoying amount of time.
> 
> Maybe launching the getty should shut off boot messages instead. Or maybe
> this should happen after a configurable IdleTimeout instead of having
> Type=idle always wait until end of transaction.
You mean like the code in execute.c does?

#define IDLE_TIMEOUT_USEC (5*USEC_PER_SEC)
...
r = fd_wait_for_event(idle_pipe[0], POLLHUP, IDLE_TIMEOUT_USEC);

if (idle_pipe[3] >= 0 && r == 0 /* timeout */) {
      /* Signal systemd that we are bored and want to continue. */

;)

Zbyszek