[systemd-devel] Hotplug auto mounting and masked mount units

[Lennart Poettering]

On Do, 09.01.20 15:56, Phillip Susi (phill at thesusis.net) wrote:

> Someone in #debian mentioned to me that they were getting some odd
> errors in their logs when running gparted.  It seems that several
> years

gparted really shouldn't mask units, that's just wrong.

They should just take BSD file locks, as documented here:

https://systemd.io/BLOCK_DEVICE_LOCKING

> ago there was someone with a problem caused by systemd auto mounting
> filesystems in response to udev events triggered by gparted, and so as a
> workaround, gparted masks all mount units.  Curtis Gedeck and I can't
> seem to figure out now, why this was needed because we can't seen to get
> systemd to automatically mount a filesystem just because it's device is
> hot plugged.  Are there any circumstances under which systemd will mount
> a filesystem when it's device is hotplugged?

This is now controlled by the x-systemd.device-bound mount option, see
here:

https://www.freedesktop.org/software/systemd/man/systemd.mount.html#x-systemd.device-bound

> Also I'm pretty sure this part is a bug in systemd: any service that
> depends on -.mount ( so most of them ) it will refuse to start while
> -.mount is masked.  It shouldn't matter that it's masked if it is
> already mounted should it?  Only if it isn't mounted, then it can't be
> mounted to satisfy the dependency.

Can you file a bug about this? Sounds like something to fix.

(But really, don't mask -.mount, really don't, masing is a heavy heavy
hammer, and not appropriate for clean codepaths)

Lennart

--
Lennart Poettering, Berlin