--- /lib/systemd/system/ssh.service.klaux.20170222 2017-02-22 12:14:44.669510431 -0300
On Sun, 02 Nov 2014 11:13:47 -0800 Russ Allbery <
rra@debian.org> wrote:
> LACROIX Jean Marc <
jeanmarc.lacroix@free.fr> writes:
>
> > I confirm this behaviour on Debian Jessie with following packages....
>
> So far, everyone who has had this problem is having trouble because they
> have mounts configured in /etc/fstab that are failing. Judging from the
> large number of mounted file systems you have, I suspect a similar issue.
> systemd takes a different attitude towards failing mounts and treats them
> as fatal by default.
>
> If you mount file systems from /etc/fstab that may or may not be present,
> could you try adding "nofail" to the mount options for those mounts?
>
> This is really a bug in the process of switching to systemd that doesn't
> catch this problem and either add nofail or provide some warning. We need
> to sort this out before the release.
>
> However, separately...
>
> > In order to be sure, i make a trivial test .....
>
> > root@choobaka:~# /usr/sbin/sshd -t; echo $?
> > Missing privilege separation directory: /var/run/sshd
> > 255
>
> > which is a correct behaviour for sshd daemon.......but ....
>
> > root@choobaka:~# /etc/init.d/ssh start;echo $?
> > [ ok ] Restarting ssh (via systemctl): ssh.service.
> > 0
>
> > ..is not a correct behaviour because systemd can NOT detect error.
>
> > It seems that /etc/init.d/ssh script does not test output value for sshd
> > daemon
>
> ...this sounds on the surface like a good idea. In general, I think any
> daemon that has a configuration and a test mode should test the
> configuration before starting, or at least before restarting, so that it
> can clearly diagnose configuration errors. Maybe I'm missing some problem
> for doing this with sshd, though.
>
> --
> Russ Allbery (
rra@debian.org) <
http://www.eyrie.org/~eagle/>
>
>