Re: '/dev' nodes not getting populated

[Vibhav Sharma <vibhav@drishti-soft.com>]

Hi Michael,

   Well, I was the on who originally reported the problem. On my 
system, the problem seemed to be that the 'initscripts' configuration 
had been messed up.
I compared mine with a friends whose system was working and noticed the 
anomalies. I used 'sysv-rc-conf'. 'udev' is supposed to run at 'init' 
mode 'S' and that entry was unset. Setting that solved the problem. I do 
not know how and why it happened so can't help you there. Just check and 
see if this is the problem. Couple of other services configuration was 
also messed up.

  What you would need to check I think is that 'makedev' runs at 'init' 
mode '2-5', and 'udev', 'udev-mtab' are set to run at 'init' mode 'S'.

  Hope this helps :). And thanks to Kevin who pointed out that it was 
an initscript error :).

-Vibhav


Michael Armida wrote:
> Hello All,
>
> I picked up on a recent conversation from this list here:
>
> http://permalink.gmane.org/gmane.linux.debian.user/256763
>
>
> I'm having exactly the same problem.  Does Kevin Mark or anybody else 
> know where I can find more information about the "initscript upgrade 
> error", or how to fix this problem?  I checked around in 
> bugs.debian.org/initscripts, bugs.debian.org/sysvinit, and google, and 
> can't find anything useful about the situation.
>
> A friend and much wiser Debian user than I took a look at my problem 
> and determined that by commenting out the "return" in the 
> mount_tmpfs() below from the udev startup script, /dev gets populated 
> as normal, and things go back to usual, minus the hacked script 
> (around line 23):
>
> mount_tmpfs() {
>   if grep -E -q "^[^[:space:]]+ /dev tmpfs" /proc/mounts; then
>     echo "Looks like /dev is already mounted.  Let's keep going anyway."
>     # return
>   fi
>
>
> We weren't, however, able to determine why /dev was being mounted with 
> the tmpfs ahead of the udev script.
>
> I am but a newbie Debian user, so please feel free to point me at the 
> right documentation or bug report.
>
> TIA,
> Michael