Hello James, On Thu, 11 Dec 2008, James Antill wrote: > The main reason¹ it was a daemon initially is that pirut/pupplet spoke > to it over dbus. > Changing yum-cron to use yum-updatesd --oneshot was on my TODO list a > long time ago, but it turned out to be non-trivial and keep backwards > compat. with both pieces ... so nothing happened to either piece. well, we all know nowadays, that it isn't really perfect, if everything depends on dbus. Yes I know, I'm rubbing salt into the wounds. But stuff is also breaking very easily. > ¹ One other feature is that we wanted "hourly" but with a random start, > so if everyone turned their computer on at 9am it wouldn't slam the > repo. ... this might be possible in newer cron's though. This is already possible with every cron: "sleep $(expr $RANDOM % 7200)" is something, spamassassin already uses inside of the cronjob. That could have been lightly adapted for a yum cronjob to generate more randomness and to enlarge the possible timeframe. Of course I understand, that dbus is nice and so on, but I'm not seeing how it is really useful. Again, the push happens about every 24 hours and the cache of the downloaded files by yum expires after maybe one hour, if I am not wrong here. Why do we generate such unnecessary load and traffic? Greetings, Robert -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
