Bugzilla – Bug 153570
Problems after update
Last modified: 2006-04-27 10:44:02 UTC
I just updated my openSUSE 10 system, and it no longer boots properly. It can no longer access my network, and never completes runlevel 5. I can boot into runlevel 3, but if I try for runlevel 5 it just hangs after setting up SSH. It does respond to the keyboard, but fails to start X. The network ... as you can see, it works fine if I boot a different system, but not openSUSE 10. :( I had to boot FC3 to submit this report. This is with the latest kernal (2.6.13-15.8-default) and all other applicable updates identified by YOU.
What do you mean by "updated" - did you just installed some new patches via YOU? Please attach the log files from your suse system: http://en.opensuse.org/Bug_Reporting_FAQ#YaST
To find out what is causing the problem, turn off parallel booting: /etc/sysconfig/boot:RUN_PARALLEL. Use the rescue system or enter 1 as a boot kernel parameter (for runlevel 1) to be able to edit the file.
I can boot into runlevel 3 fine, and from there start X, but for some reason it just won't boot to 5. The network issue seems to have resolved itself, not sure why it was failing. I inherited another computer but still only have one monitor, I'll connect the older system back up over the weekend as see what I can figure out.
Just some other details, before I see what disabling run parallel does. I had in fact just installed the latest versions of Gnome, KDE, and Firefox when I decided that I should check for updates. I gather your SUSEwatcher icon under KDE doesn't actually do anything on its own, there were quite a number of updates ... I have what may be a fairly unique system here. I have a 20 GB main drive, and a 40 GB secondary, on a 500 MHZ K6-II with 448 MB RAM. I used to be an alpha tester for Mandrake, at one time I had 4 completely different versions of Mandrake (and Windows 98 SE) installed simultaneously. Due to my BIOS being unable to boot a 40 GB drive, the secondary is invisible to Windows, and every system I install has to have at least /boot on hda. So in fact, /boot is shared by several versions of Linux and is hda5. Mandrake 10 (or really, using the 2.6 kernel) had mishandled a BIOS extender that came with the 40 GB drive, rendering the other partitions unbootable at the time. (I was able to fix the filesystem identifier to get Windows bootable, but since 2.6 hadn't recognized the partition type on my original /boot, all my prior kernel versions were lost.) When I finally decided to install a newer version of Linux, I had DVDs for FC3 and openSUSE 10 from a magazine, I decided to try both which meant recreating /boot as a separate shared partition. FC3 is on the primary drive, openSUSE (other than /boot) is on the secondary. Since Mandrake (now Mandriva) defaults to LILO, I have not reinstalled it on this computer though my other computer now has Mandriva 2006 on it. Anyway ... so currently my GRUB menu lists FC3, openSUSE 10 and Windows 98 SE, and other than the shared /boot everything is independent, and ran well up to this last update. (Since I prefer the KDE file manager, I really don't use Fedora - too many of their GUI utilities are oriented towards Gnome.) But that's the nature of the beast I'm having this trouble with, and prior to the last update everything worked as expected. It is actually sometimes still telling me that the network logon failed, as you see the network started fine. (Oh, I share a LAN with my dad, the network is through a router which supports DHCP. Even though I only have one network cable to this room and so swap it between computers, this is a LAN network.) Well, let's submit this, then see what happens with the non-parallel boot.
Well, the last line displayed is "Starting D-BUS daemon done", nothing gets displayed following that. Which should mean the problem is there or in the next step, checking /etc/init.d/rc5.d it looks like the next step is S03earlykdm. No line is displayed indicating that step, though, so I can't determine for sure whether earlykdm was called or not. On the other hand, since runlevel 3 works fine and does contain s03dbus (and of course doesn't contain s03earlykdm) it seems logical that the problem does lie somewhere in earlykdm.
Conclusion: checking on the SUSE KDE mirror, I see an update to kdebase-kdm. Install it, my booting problem is gone. Would be nice if the system had provided enough info to isolate the problem, rather than a step failing with no feedback to indicate which step had failed. Still the simply cosmetic bug with network appearing to fail randomly (key word - appear) while everything continues to operate properly, which turned out to be a distraction from the other problem. As far as the failing to boot completely, that is resolved. What you will do about the lack of feedback in earlykdm and about the spurious network failed messages is up to you guys.
Martin, any comments?
Steve, check the kdebase-kdm changelog to find the KDM bug and continue there. Concerning the network problem, it looks a bit like bug 140124. Reassigning to sysconfig.
As the kdm bug has already been fixed, I see no reason to pursue that. The supposed error in network ... I don't know whether to pursue it or not. My system works, I have no problems connecting to the internet, but during init the system indicates that network failed. A timeout while waiting for DHCP is usually the message when I see a specific message, but sometimes I just see "failed" with no details. Your call.
Can you send me the lines from network start in /var/log/boot.msg when it failed? And even if network start script said failed it works later? Please also attach 'hwinfo --netcard' and all /etc/sysconfig/network/ifcfg-* files (except 'lo').
Seems there is nohing to do anymore.