I came to this list via Google while trying to find some information
after having a rather unpleasant experience recently: I was engaged in
a last-minute professional email exchange trying to beat a critical
deadline while on a flaky tethered wireless connection when something
(I think it was packagekitd) started downloading megabytes of
presumably package metadata, almost completely saturating the link.
I desperately tried to kill it, but maybe systemd respawned it
immediately, at least I couldn't figure out how to do it and I
couldn't Google for help as I usually do as the link was - well -
I was lucky the whole thing finished in time, but now I am trying to
fix the issue for good. I found that it has been discussed in passing
as a side topic under the subject heading "F21 downloads repository
metadata in 3 places!" without really coming to a resolution, and I
found bits and pieces of information that leave me somewhat confused.
I believe that this is a critical issue that users should not fight
for themselves, but where the distribution should assist and be
Here are some specific points for comment and consideration:
First, I am very sure that F20 did not do this, it must have happened
quietly after I did a fedup update (which otherwise worked well).
Second, I have some trouble understanding how the pieces of software
fit together. I am a technical user and not afraid to dig into
details if necessary, but this is more complicated than I expected it
to be. There is yum, dnf, packagekitd (I saw it with top), and Gnome
software potentially involved, but instructions I find on the web seem
to be ambiguous who is responsible for automatic metadata downloads.
Some instructions point me to open the Gnome software dconf settings,
where there is a "download updates" option (how about metadata then -
I find it unlikely that actual packages were downloaded because the
complete download volume was in the several MB range). Also, there is
a listing "compatible-projects" which contains Gnome, KDE, XFCE. But
I am a happy cinnamon user. Does this apply to me?
I am sure I can find all of this out given enough time, but I think
that if I am finding this difficult, there will be many others who
will be in a similar boat, and there should probably be some
brainstorming at the distribution level on how to set/unset such
actions, and also how to protect users from sudden unexpected behavior
which can render the network practically unusable at random times.
Last, in the mailing list thread mentioned earlier, I read the opinion
that nowadays bandwidth is not as much an issue as it used to be. In
my experience, also from observing how others use their devices, this
is not the case: we are now using mobile networked devices in
situations where a few years back we would not have dreamed being
online, including on airplanes, in remote spots with flaky mobile
reception, or at crowded conference venues where everybody is
competing for bandwidth on few configured access points. And we are
expecting that we can get work done this way. Thus, in my experience,
the ability to control who is using bandwidth at any time is a crucial
capability of an operating system and should be exposed to the user in
a more explicit way than fiddling with dconf settings.
Show replies by date