Comment # 5 on bug 1187289 from Franck Bui

(In reply to Rodrigo Valentim from comment #4)
> For systemd I guess that's a too small issue to create a validation only for
> that, maybe just a more accurate message would be interesting, if it's not
> too much work.

systemd is supposed to inform you while it's waiting for the relevant device to
appear:

   [***   ] A start job is running for /dev/xxx (1min 3s / 1min)

Then you should see something like the following when the timeout expires:

  [ TIME ] Timed out waiting for device /dev/xxx.
  [DEPEND] Dependency failed for /mnt.
  [DEPEND] Dependency failed for Local File Systems.
  [DEPEND] Dependency failed for Early Kernel Boot Messages.

Once you get the emergency shell, you can still look at how it failed by
inspecting the content of the journal with `journalctl -b -p warning`.

> But for the "auto mount" process, despite of be a small issue, should be
> quite easy to add the "nofail" option, no? 

Well I think the question is more about how your USB stick ended up in fstab.
The component who did that should consider whether the device is really needed
during the boot process or not and add "nofail" (or probably the better
alternative "x-systemd.automount").