/usr/lib/systemd/system/dbus-org.freedesktop.Avahi.service is installed by the avahi package. /etc/dbus-1/system.d/avahi-dbus.conf defines, that the above service is used to start /usr/share/dbus-1/system-services/org.freedesktop.Avahi.service, which in turn starts the systemd-service. This all happens without any user interaction which is why my synergy, someday pulled in avahi unnoticed, which was then started automatically and I was humiliated by my friends for having a running avahi. Please let the user explicitly has to enable the service via systemctl, so that it can't automatically be started just via dbus after (maybe unwanted) installation. Reproducible: Always
Pacho: This seems to be your doing. I would recommend reverting 80f3f8ce6f234b51254947dc0f579fe843972961. You can add a postinst message advising people to enable it if they want.
(In reply to Mike Gilbert from comment #1) > Pacho: This seems to be your doing. > > I would recommend reverting 80f3f8ce6f234b51254947dc0f579fe843972961. > > You can add a postinst message advising people to enable it if they want. let's try to post patches for review rather that directly committing except for trivial stuff. i'd like to keep a abreast of what's happening with avahi.
(In reply to Anthony Basile from comment #2) My suggestion would be to simply remove the symlink creation from src_install. Bug 537000 should probably have been a WONTFIX or INVALID.
(In reply to Mike Gilbert from comment #3) > (In reply to Anthony Basile from comment #2) > > My suggestion would be to simply remove the symlink creation from > src_install. Bug 537000 should probably have been a WONTFIX or INVALID. Not so simple because some ebuilds are now stabilized. I added -r11 to the tree dropping the sym link and dropping everything to ~arch. I'll wait about a week or so for testing and then autostabilize all the arches since this is a trivial change. @systemd folks, please test to make sure this is what you want.
I would really like Pacho's feedback as the one who made the change in the first place. I'm approaching this as a systemd user who doesn't use avahi; he may have another perspective.
I agree with Jonas here. Avahi is a network service, and I would rather not have those started automagically without explicit 'on' switch just because something pulled libavahi in. Not sure if running avahi could actually cause any security issues but it discloses information you may not wish to disclose. And I'm systemd user who actually uses avahi.
(In reply to Anthony Basile from comment #2) > (In reply to Mike Gilbert from comment #1) > > Pacho: This seems to be your doing. > > > > I would recommend reverting 80f3f8ce6f234b51254947dc0f579fe843972961. > > > > You can add a postinst message advising people to enable it if they want. > > let's try to post patches for review rather that directly committing except > for trivial stuff. i'd like to keep a abreast of what's happening with > avahi. I waited for months for any reply in bug 537000
I reported it to upstream when avahi was "revived" and we are a bit lost about how systemd people are expecting us to deal with that stupid situation in that systemd silently fails to start avahi when needed (for example, every time I try to print something from GTK+ dialogs, it tries to load zeroconf exported printers and you will see that they are not shown for some "unknown reason" until you see in journal logs the systemd errors of it trying to start avahi and failing): https://github.com/lathiat/avahi/issues/29 If anyone wants to help there, please do :)
(In reply to Pacho Ramos from comment #7) > (In reply to Anthony Basile from comment #2) > > (In reply to Mike Gilbert from comment #1) > > > Pacho: This seems to be your doing. > > > > > > I would recommend reverting 80f3f8ce6f234b51254947dc0f579fe843972961. > > > > > > You can add a postinst message advising people to enable it if they want. > > > > let's try to post patches for review rather that directly committing except > > for trivial stuff. i'd like to keep a abreast of what's happening with > > avahi. > > I waited for months for any reply in bug 537000 The bug simply quotes the Arch Linux "fix" without any proposed course of action. I cannot fix/test systemd bugs.
What is the behavior of avahi when NOT running systemd? Does it get auto-started in response to a dbus request?
(In reply to Mike Gilbert from comment #10) > What is the behavior of avahi when NOT running systemd? Does it get > auto-started in response to a dbus request? it does not
(In reply to Mike Gilbert from comment #3) > (In reply to Anthony Basile from comment #2) > > My suggestion would be to simply remove the symlink creation from > src_install. Bug 537000 should probably have been a WONTFIX or INVALID. since the symlink creation has been reverted, this bug should be fixed