Bug 79103

Summary:	udev mount not working in systemd-212-3 (works in systemd-210-3)
Product:	systemd	Reporter:	neodampf
Component:	general	Assignee:	systemd-bugs
Status:	RESOLVED WONTFIX	QA Contact:	systemd-bugs
Severity:	normal
Priority:	medium
Version:	unspecified
Hardware:	x86-64 (AMD64)
OS:	Linux (All)
Whiteboard:
i915 platform:		i915 features:

Description neodampf 2014-05-23 05:49:45 UTC

I don't know what got changed concerning any udev/mount code between systemd-210 and systemd-212, but I noticed some inclusion of a new library (libsystemd) and a message about some new stuff added to udev for systemd. 

Well, whatever the systemd team did, mount'ing and umount'ing usb drives with udev rules stopped working(no effect) AFTER I upgraded systemd to the most recent version(212). 

That and another issue I didn't have a chance to (or am too lazy to) diagnosis if it was caused by this as well, but unmounting w/ umount was not causing any kind of I/O sync at all(just typing umount in console).

e.g. copy a 57mb wav file to usb, unmount it, take it out, plug it in,remount and then the wav file was 0mb.

Comment 1 Kay Sievers 2014-05-23 21:46:23 UTC

This is the expected default behaviour, udev running on systemd systems
cannot mount volumes system-wide:
  http://cgit.freedesktop.org/systemd/systemd/tree/NEWS#n97
  http://cgit.freedesktop.org/systemd/systemd/commit/?id=c2c13f2df42e0691aecabe3979ea81cd7faa35c7

Comment 2 neodampf 2014-05-28 14:50:01 UTC

ok hang on, what about umount ?
Are you saying that it's not supposed to flush I/O operations ?

Comment 3 Kay Sievers 2014-05-28 16:00:43 UTC

(In reply to comment #2)
> ok hang on, what about umount ?
> Are you saying that it's not supposed to flush I/O operations ?

Flushing to a device that is removed? But we don't really know how
to do that. :)

Udev is and never was capable to properly handle mount operations, if you
really need that and know what you are doing, customize the config of the service file.

The default setup cannot and should not support anything like that.

Comment 4 neodampf 2014-05-28 21:46:52 UTC

Really sorry I'm afraid my posts regarding umount was ambiguous.
I wasn't referring to making udev call umount.

I was referring to running umount (e.g. from terminal) not working as a separate isue.

From the first post I made I paired the udev & umount problems as related issues and that must have caused the confusion that I was talking about umount called from udev rules.

Please reread my posts about umount in omission of the udev context.

Comment 5 neodampf 2014-05-28 22:29:57 UTC

Ok I know what happened now.
I get that the latest update of systemd took out the original functionality that enabled udev to mount/unmount drives from calling scripts. But my old udev rule was still being read. Apparently it had no easily detectable effect on the mount status of any drives, but its getting called still messed up manual mount/umount somehow.

To me, that's still kind of buggy in nature, I get that the intention was to disarm udev from  mounting, but now the effect of it getting called from udev is like a hidden error. 

If the new desired behavior for udev mounting is to do nothing shouldn't it be rendered completey silent and harmless then ? Shouldn't it be implemented that way rather than some buggy consequence that doesn't really make itself known to me unless I look at my file sizes carefully ?

Comment 6 Lennart Poettering 2014-06-23 16:21:58 UTC

Well, we use what is available. And mount namespaces are simple and work. And I see no other nice way to achieve the same.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.