On Thu, 2006-08-17 at 06:21 -0400, Phil Schaffner wrote: > On Thu, 2006-08-17 at 09:16 +0100, Philip Wyett wrote: > > If it was a server where uptime was important then use of telinit > > would > > be appropriate. But not many folks install the proprietary nvidia > > driver > > on servers and for workstations or laptops reboots don't really > > matter. > > Describing it as bad practice is ridiculous. > > Very nice howto with the exception of the reboot advice. It is somewhat > short of ridiculous to describe unnecessary reboots as bad practice > IMHO. Newbies would be well served to see that one of the [many] great > advantages of Linux over Redmond OSs is the absence of the perpetual > reboot cycle. Rebooting for every change is way too windows-like for my > taste, and I have workstation (and server) machines that regularly go > months without a reboot in the absence of kernel updates. > OK ... lets look at this objectively. You are usually installing an Nvidia driver because: 1. You just turned off the computer, installed a video card, and just turned on the computer. 2. You just installed CentOS for the first time. 3. You just installed a new kernel. In either case ... you just booted the machine. It has 25 seconds of uptime. Another reboot is NOT going to matter. It is certainly best practice, in my opinion, to reboot your machine if you need to restart X. You certainly can do telinit 5 after to did telinit 3 ... but I always reboot and I always recommend rebooting. If you have a reason not to reboot, fine, but IMHO it is a very good practice to reboot ... if for no other reason than to make sure that it works and there is not a problem with udev creating the nvidia device, etc. If there is a problem, you want to find it now ... while you are installing the driver, and not after a power loss when you reboot needing to access a very important presentation you were just getting ready to use. > > > > In the mail I clearly mentioned telinit could be used but did not > > expand > > on it, just specified I personally did not use it as is my choice. > > No problem with your choice - Linux is all about choice - and you > clearly covered the option most of the responders seem to prefer - just > not clearly explained. > > One other viable choice IS to use 3rd party repos, which may be easier > for a newbie as well as being consistent with frequent advice to use RPM > packages on an RPM-based system. I have had very good experience with > ATrpms nVidia kmdl packages; although I'd recommend against using Axel's > packages wholesale for a stable CentOS system. The yum includepkgs > directive is quite useful here. > > > Should I spend more of my time writing all the possible options just > > to > > make some people happy - Not a chance, I've got plenty of other non > > and > > CentOS related work to do! > > I hear you. Perhaps someone can find time to build on this for a WiKi > article. Will take a crack at it if I can find the time. :-) > That also says to reboot :)
Attachment:
signature.asc
Description: This is a digitally signed message part
_______________________________________________ CentOS mailing list CentOS@xxxxxxxxxx http://lists.centos.org/mailman/listinfo/centos
