hmm, on Thu, Sep 17, 2009 at 11:43:07AM +0200, Stephan A. Rickauer said that > Our Institute moved away from Linux servers always everwhere, just > *because* of updates are unreliable. Very often we did an apt-get update > or an yum bla, reboot, machine dead or fucked up otherwise.
everyone is comparing apples and oranges here. linux is a bunch of packages. openbsd is base (and then probably some packages). when updating linux "the OS", one is still updating packages. openbsd, the system, is clearly easier and more consistent to upgrade. but updating packages on openbsd is more time consuming than on e.g. linux. so yes, when 4.6 comes out, i'll update a server in maybe 15 minutes with all its packages as well -- because 4.6 will come with packages. but if i were to update a package with a lot of dependencies in say 3 months because it has a vulnerability or reliability fixes, then i have to do the package dance myself. depending on the package this might be easy, or it might be hell. but it clearly takes more time and effort than in linux, this team just doesn't have the manpower to compete with that. and if you have a handmade inhouse solution to roll out a package like that for all your 1000 machines, great, you are earning your money as an admin. but calling people names because they are using an update infrastructure in place seems juvenile to me at best. bind was as special example because in linux it's just a package, and while it might be in openbsd as well, it is provided in base. and that brings up the theme of binary patching, and the archives are full of it. -f -- "fishing, stranger?" "no, just drowning worms."