On Sun, Oct 28, 2018 at 05:43:28PM +0100, Jayesh Badwaik via arch-general
I am looking at possibilities to have a setup which does not need to be restarted at all and can be live patched. I have seen this old post:
I think that if your system is important enough that it shall never go down even for a few minutes, what you really want is additional redundancy in form of a high availability setup. Your hardware and software can fail at any time and then it doesn't matter if you can hotpatch your kernel or not. If you have a HA system, it doesn't matter if you reboot or not because the other node(s) take(s) care of the work for the time being. Also sometimes the power fails and if you don't reboot your machines regularly, they may not come back on after that (ask any bigger data center). If you reboot regularly, chances are higher that you spread out any problems and resolve them before all machines are rebooted at once. If you want to do it for science, all I can say is that I think there are more interesting/important problems than this, but YMMV. Well, and then there's the issue that you need to specifically create the patches and that you can't patch everything because that's just way too much work, but that has already been mentioned by others in this thread. Florian