Re: Btrfs by default status updates, 2020-07-26

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mo, 27.07.20 19:20, Tomasz Kłoczko (kloczko.tomasz@xxxxxxxxx) wrote:

> Really?
> (my jaw is on the floor crashed into the pieces)
> I must honestly confess that in the meantime systemd swallowed autofs as
> well!

It didn't. It supports a small subset of the whole autofs logic, in
order to make fs-based activation work, i.e. provide a path in the fs
before its mounted in order to improve robustness, increase
parallelization and pull things in on demand only.

i.e. a good usecase for this is mounting the ESP with this, so that
the fs remains untouched unless actually accessed. i.e. for file
systems that are fragile to some point and which are seldom accessed
it's great to guarantee that they are clean (i.e. not dirty/mounted)
whenever possible. And when it is accessed we pull in fsck@.service
for the fs in the bg and then mount it, all invisibly to the
program accessing it.

Another great usecase for this is actually any form of removable
media, such as USB sticks: it allows us to pretend-mount removable USB
sticks without actually having the fs mounted in the kernel all the
time. This means we have a much larger chance of leaving a clean fs on
the USB stick if the user yanks the stick from the plug without
cleanly unmounting it first, because the fs is simply not mounted at
all unless accessed.  systemd-mount exposes this nicely (and ideally
udisks would always use this functionality for all USB media, which
would also mean fsck would be called at the right times, which udisks
currently doesn#t call at all, which sucks hard).

What's also great is that we can properly detach file systems in the
right order if systemd knows the deps. i.e. a service backing a mount
can be stopped at the right time when an autofs mount is shutdown,
which is hard if the stuff is entirely opaque to systemd...

BTW, this was one of the features available pretty early on in
systemd, since before systemd was even adopted by Fedora. No need to
lose it about this so many years after. And it's not complex code,
it's a relatively simple API (though it feels a bit dated in style).

It's not usable for NIS stye autofs mounts, as it doesn't support
maps. Use autofs4 package if you want that.

Lennart

--
Lennart Poettering, Berlin
_______________________________________________
devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx




[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Fedora Announce]     [Fedora Users]     [Fedora Kernel]     [Fedora Testing]     [Fedora Formulas]     [Fedora PHP Devel]     [Kernel Development]     [Fedora Legacy]     [Fedora Maintainers]     [Fedora Desktop]     [PAM]     [Red Hat Development]     [Gimp]     [Yosemite News]

  Powered by Linux