Bug#852696: gvfs-fuse: automatically starts gvfsd and gvfsd-fuse in root user slice

[Simon McVittie]

Control: reassign 852696 appstream

On Thu, 26 Jan 2017 at 19:27:19 +0100, Laurent Bonnaud wrote:
> janv. 26 13:55:02 hostname dbus-daemon[16957]: [session uid=0 pid=16957] Activating via systemd: service name='org.gtk.vfs.Daemon' unit='gvfs-daemon.service' requested by ':1.1' (uid=0 pid=16956 comm="appstreamcli refresh-cache ")
> 
> Do you think I should reassign this bug to the appstreamcli package?

I think this is an appstream bug, and I've reassigned it.

(Context for the appstream maintainer: Laurent found that various
gvfs services were unexpectedly running in root's `systemd --user`
session.)

I'm guessing this is caused by /etc/apt/apt.conf.d/50appstream having
this? Had you run 'apt update' in your root shell shortly before that
activation happened?

APT::Update::Post-Invoke-Success {
    "if /usr/bin/test -w /var/cache/app-info -a -e /usr/bin/appstreamcli; then appstreamcli refresh-cache > /dev/null; fi";

I think either appstreamcli's main(), or that apt configuration snippet,
should set the various magic environment variables to make GIO behave
less like a desktop and more like a system facility - I assume appstreamcli
doesn't actually need gvfs etc. to do its job! See also
<https://bugzilla.gnome.org/show_bug.cgi?id=767172>,
<https://bugzilla.gnome.org/show_bug.cgi?id=767183>.

This particular one would be solved by "GIO_USE_VFS=local", I think.

GIO_USE_VOLUME_MONITOR=unix and GSETTINGS_BACKEND=memory might also
be worth considering, although I suspect appstreamcli won't use
either of those in practice.

    S