Hallo, * Osamu Aoki [Mon, Apr 05 2021, 08:32:30PM]: > Hi, > I created a script to check all broken repo contents across all debian > and security downloads. > > After fiddling with this problem, I realized few things on this apt- > cacher-ng problem. > > Broken downloads is not just deb file but I also saw truncated > InRelease file. But such problem is much fewer than simple missed > downloads. I see them in log.
Hm, thank you for the verbose report. All the things you reported make me worry a lot. Yes, it sounds very much like 954904 all over again. For a while it did happen to me too, but it's a Heisenbug, I never managed to reliably catch it and after some refactoring it disappeared. I never trusted that outcome, thus 954904 is still open. If you don't mind, I would like to merge them. > So after fixing apt-cacher-ng data with attached script, running apt > again goes smoothly. Yes, but the question is for how long. I am preparing a bigger upstream release right now, where the logics of file storage are overhauled to a large extent. This might solve your issue along the way. Especially, there won't be truncation of payload anymore, therefore no risk of getting into such trouble by design. If you have a good environment to reproduce that issue, would you volunteer to test it in a couple of weeks? (or maybe even sooner, cannot tell yet) > As I read over BTS, I saw interesting one on approx BTS. Debian Bug > report logs - #884713 > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=884713 Under > systemd, socket activation is rate limited. > > This socket activation limit may be the root cause. What do you think? See /lib/systemd/system/apt-cacher-ng.service , socket activation is not used by default. Also it wouldn't explain sudden trancation in the cache. Best regards, Eduard. -- <the-me> gibts fürs debian pakete erstellen nicht auch ne GUI lösung?
