Well, I ran urpmi yesterday – partly to see what would come down the pipe, and partly out of hope to fix both a nagging problem with being unable to boot new kernels and a new problem induced since the last update that prevented USB devices from working after about a minute.
Of course, I only discovered the new one after I desperately needed to copy some stuff over to a USB thumbdrive… (Figures.)
So I updated in run-level 3, and was nonplussed when I was unable to go to any other run-level after the update completed. Additionally, at some point during the update, the network interface was nuked, and I had to run “service network restart” in order to get eth0 visible again – running “ifconfig eth0 up” would fail, squawking that it no such device could be found. Once networking was working again, I reran the urpmi command and finished updating with only minor drama. Until I finished, reinstalled my nvidia kernel driver (ver. 9764), and typed “init 5” in order to bring up the desktop. At this point all I got was a cryptic message that there were no init processes left, or words to that effect.
So, I rebooted, cuz it wasn’t yet enough drama and I was bored.
So, it failed to boot past a new message to enter the init level desired, instead always responding that there were no init processes left (or words to that effect).
So, I booted off my trusty RIP (Rescue Is Possible) liveCD, mounted my boot partition, copied over a working mdadm.conf to /etc/mdadm/mdadm.conf, followed instructions I left myself in the file, and soon had my RAID and LVM volumes up and mounted. I checked the mounted /etc/inittab, and it had next to nothing it it. Luckily, a backup existed, called inittab.rpmsave, which I copied back over the impaired inittab file. I also confirmed that the update had wiped all of the files from /etc/rc0.d through rc6.d directories. I unmounted the volumes and rebooted to see if the system would come up anyway.
It did, in the default run-level 3, but init # would not work. So I ran gdm to start the desktop and it works fine. I am guessing that Mandriva Cooker has been changed to move away from the current initscript process to a different one (sysV to BSD-style, maybe?). I understand that Slackware uses BSD-style initscripts, so maybe this is what is going on. Unfortuately, I am not even remotely familiar with this, so I will start studying.
Good news is, I got KDE 3.5.6, the USB problem is fixed, and the kernel boot problem I have had for a very long time (never found a cure, but am sure it was a Cooker-related problem) is gone. As soon as kernel 184.108.40.206 comes out (the git-6 changelog seems to indicate a lot of SATA and IDE changes), I will throw that on and see how it works. Beryl also got updated, and some quirks with the old version are gone. Nice.