Public bug reported:

Binary package hint: gvfs

The "Places" menu in Ubuntu Lucid  is populated with inconsistent entries, or 
rather it is overpopulated.
The issue seems to be RELATED to bug #442130, but it's not about duplicate 
entries but rather wrong entries.
Encrypted local volumes and network (nfs) filesystems are problematic.
1). filesystems on encrypted block devices (using luks)
I set up /home as an encrypted filesystem. I use sda3 for that. I put esda3 
/dev/sda3.... in /etc/crypttab and 
UUID=xxxxx......    /home           ext4 ..... in /etc/fstab. UUID is the uuid 
of the encrypted filesystem, not of the underlying encrypted block device. 
Everything works fine, the system asks the passphrase for the encrypted volume 
at boot and then i can log in.
However, in the places menu an entry is added under removable media for "esda3" 
- which eventually could be ok only if one could use it : if it is clicked an 
error apears with :
-------------------------------
Unable to mount esda3
/dev/mapper/esda3 is mounted
-------------------------------
I should mention that I get the same error if I put /dev/mapper/esda3 instead 
of UUID=xxxx in /etc/fstab. Somehow nautilus/gvfs is not able to detect that a 
certain entry in fstab refers to an already mounted filesystem and will try to 
mount it. The kernel however knows you are trying to mount an already mounted 
resource and so the mount  does not happen and we get that error.
2). The same as above happens with nfs filesystem entries in fstab that were 
previously mounted using CLI.
For example an entry like :
host:/mnt/d5        /media/d5      nfs     user,noauto,tcp,hard    0       0
will be added to the places menu in both the system menu, nautilus and file 
open dialogs. However, if one mounts that filesystem using CLI, trying to 
access that fs using system menu's  places  fails as (again) nautilus tries to 
mount the filesystem by itself, not realizing that the fs is already mounted. 
And of course this ends with another mount error instead of just 
entering/browsing that resource.It seems however that opening a new nautilus 
window will correctly see those filesystems as mounted and does not try to 
mount it again. So for nfs we get the mount error only with the system menu 
places submenu but not with the "computer browser" (nautilus) or open dialog. 
This however is NOT true for the encrypted filesystem (number 1, above). In 
that case you get the mount error in all the scenarios.

3) Finally, one could have quite a number of transient filesystems in 
/etc/fstab and only a few to be active at a time. It is not a sane default to 
put all these fstab entries in the places menu, as it really clutters both 
nautilus and the open file dialog with mostly not used entries. Instead it 
should only show already mounted filesystems, OR it should be possible for the 
user to hide some entries.
For example we could have a bogus mount option in fstab like "gui_ignore" or 
something like that.
$ cat /etc/lsb-release 
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu 10.04 LTS"

ii  gvfs                                        1.6.0+git20100414-0ubuntu1      
            userspace virtual filesystem - server
ii  nautilus                                    1:2.30.1-0ubuntu1               
            file manager and graphical shell for GNOME

PS : if there is any way to control at least the hiding of the
problematic entries that would be very useful, but i could not find any
docs about how gvfs/nautilus choses what filesystems to show in places
and what filesystems not to show.

Thank you.

** Affects: gvfs (Ubuntu)
     Importance: Undecided
         Status: New

-- 
Inconsistent / not functioning  "Places" entries  and nautilus tries to mount 
already mounted resources
https://bugs.launchpad.net/bugs/580684
You received this bug notification because you are a member of Ubuntu
Desktop Bugs, which is subscribed to gvfs in ubuntu.

-- 
desktop-bugs mailing list
desktop-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/desktop-bugs

Reply via email to