Re: systemd, fstab, noauto and nofail
Adam Borowski <kilobyte@angband.pl> writes:
> There's currently no way to express which mounts are needed for which
> functionality.
While that's true, I'm not sure that fine-grained control of this is
required here. We can get a long way with just a way to indicate whether
a mount is important or unimportant. And that, in turn, is exactly what
"nofail" is for, which is already supported.
The problem is that people haven't been using it, because nothing has
forced them to use it or told them that they should have been using it,
since sysvinit essentially marks every mount "nofail" in terms of the
practical effects of its behavior.
> This sounds like a failure to provide a _non-fatal_ error message.
I wish there were some way to reliably deliver to the user non-fatal boot
error messages. But for both the desktop and the server case, it usually
requires quite a bit of effort to go look at the boot messages. They
generally scroll by rather quickly, servers often have unattended
consoles, and the desktop boot messages are usually hidden pretty quickly
by the desktop manager.
> * systemd: in preinst, check if any fstab lines without noauto or nofail
> are not currently mounted -- if so, abort the installation as it would
> result in an unbootable system.
Yeah, this is the solution I've been proposing.
I also like the idea of not having ssh depend on all local file systems to
be mounted. I think it's going to be pretty rare to have a system that
has /lib and /etc mounted but can't start ssh. In theory, that's possible
with a split / and /usr, but as we've discussed in other threads, that's
an extremely unusual configuration these days.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: