On Sat, 2018-06-02 at 21:40 +0800, Ed Greshko wrote: > >>>> Seriously? This discussion comes up regularly. In most cases online > >>>> updates work, but sometimes there are issues. Read the past threads for > >>>> more info, but it's an extremely hard problem to make it always work for > >>>> very obvious reasons when you think about what's happening. > >>> I rely on tracer (using the dnf plugin) to tell me when I need to > >>> reboot. Are you saying this is unreliable? > >>> > >> Where did Sam say anything about tracer? > > He didn't. I merely state that it it's useful to know when a reboot is > > necessary and when it isn't, which is the difference between online and > > offline updates. > > I see. Then I'm confused as to your question. What are you asking about > reliability or what do you think Sam is suggesting is unreliable? Sam seems to suggest that rebooting is the only way to be sure of avoiding problems. I'm wondering if he's discounting the use of tracer as a way to do this more selectively, and if so is it because he thinks tracer is unreliable? > > > >> FWIW, I don't use the tracer dnf plugin since I use the nVidia drivers from rpmfusion > >> and akmods. It may have been improved. But when I was using it and a kernel update > >> happened the tracer process would keep the rpmdb locked so when the nVidia mods were > >> compiled they failed to install. > > I haven't had problems but I don't use the Nvidia drivers now. Back > > when I did use them I also had locking issues, though I never > > associated them with tracer specifically. > > > > > I took the time and effort to prove, at least to myself, that the asynchronous or > independent nature of the akmod process and the tracer-plugin could cause problems. > It was very variable since it depended on the number and nature of the updates. There was certainly a race condition related to locking the RPM database, so it's possible that tracer was causing this, though I don't know if anyone ever tracked it down. One wonders why it only seemed to affect akmod packages, but perhaps it's to do with the complicated scripts they run. poc _______________________________________________ users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx/message/BKFQ4CAFZ4QUY5IFKIM7EKM4ZHMQMEXW/
