[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Woody->Sarge upgrade claims to break if using devfsd



On Sat, Dec 18, 2004 at 01:21:23PM +0100, Frans Pop wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> (CC'ing d-release as this is more a release issue than an installer issue)
> 
> On Saturday 18 December 2004 12:51, Marco d'Itri wrote:
> > On Dec 18, Jeroen van Wolffelaar <jeroen@wolffelaar.nl> wrote:
> > > | udev is not configured to use devfs-style devices.
> > > |
> > > | Your system WILL BREAK if you run udev with its default
> > > | configuration.
> >
> > Actually this warning is obsolete, because since a long time udev
> > installs a devfs-compatible configuration if it can detect devfs being
> > mounted.
> 
> So, should we file a release critical bug to have this alarming message 
> removed?
> 
> > > I didn't dare to have any system rebooted yet, but fear for the
> > > results, that this upgrade caused our systems to become unbootable.
> >
> > It does not take much to RTFM and check /etc/udev/rules.d/.
> 
> IMHO this reaction is uncalled for. It may be appropriate for experienced 
> sys admins, but the average user is likely to be very concerned, and 
> rightly so, if a message like this is shown.
> Jeroen is adsolutely right to bring up this issue. Debian is _not_ only 
> for nerds.
> 
> > Actually, it's much worse to not reboot quickly after udev has been
> > installed, because it may start messing with your old /dev.
> 
> In that case, a message explaining *that* should be displayed to the user 
> instead on upgrade!

I think Marco d'Itri is trying to work on this - see his recent post to
debian-devel. He is trying to work out a correct procedure for installing udev
so that a reboot is not required.

However, on a related note I will be recommending that the user reboot their
system if possible. Since this can give a good performance benefit since some
memory will no doubtly be wasted by an already running process using an old
version of a shared library and a new process using the new one.

Cheers,

Rob
-- 
Rob 'robster' Bradford
http://robster.org.uk



Reply to: