I’ve noticed that the rate-mirrors has been picking out of date or broken mirrors more than 50% of the time I use garuda-update (which is once per day). I looked for and found the flag to stop it from updating the mirror list, but that only fixes my issues with it. I question the logic of using rate-mirrors every single time by default.
Aside from making it take longer each update (it takes more time to sort the mirrors than it saves from re-picking a “fast” mirror each time), the fact that rate-mirrors doesn’t seem to weed out out-of-date or corrupted mirrors means that more than half the time it’s run (at least in my case) it picks a bad mirror.
This is bad behavior to have on by default. Why not switch the default so that the first time it’s run it does a rate-mirror, then sets a setting requiring a flag to use rate-mirror in the future, essentially opposite how it works now. Or find a way to fix rate-mirror.
unrelated to Garuda, talking about
its defaults prevent you from using outdates mirrors, see the algorithm for details (
--max-delay options control it).
So it makes sense to either double check options, you pass to it, or share your definition of “broken” and “corrupted” mirrors.
It will often pick mirrors that have core packages at versions below that of ones that I currently have on my system (some of which haven’t been updated in days, and I only update from official repos).
One mirror in particular that it likes has a particular package that is broken in such a way that it spams the output that something is wrong with that one package about 50 times, then fails the garuda-update. Running it again to get a different mirror solves the issue.
I don’t run rate-mirror by itself, only as part of garuda update, so either rate-mirrors is terrible at determining broken and out of date repos, or garuda-update is making it so via whatever flags it uses.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.