Firefox, mutt, kicker, konsole crash after FC8 update

[Craig White]

On Fri, 2008-01-18 at 06:07 +0100, M. Fioretti wrote:
> Greetings,
> 
> this is on an AMD64 system, FC8 64bit installed with Livna nVidia
> rpms, all updates with yum applied without problems until one week
> ago, running KDE.
> 
> Last wednesday I go t the "updates available" notification on the
> desktop and told the system to upgrade. IIRC, there was some problem
> to download the evolution rpm, but I didn't bother since I never use
> it, so I told it to go ahead and update. Everything seemed to go fine,
> but then:
> 
> Firefox would crash or freeze
> Konqueror wouldn't start
> mutt ran in konsole would segmentation fault every 5/6 messages
> 
> yesterday morning, when I powered the PC up again, it failed to boot and
> left me with the prompt from which to boot and run fsck
> 
> I did, and I noticed it reported several "multiply-claimed blocks in inode"
> warnings about:
> 
> /usr/lib64/libkgroupwarebase.so.0.0.0
> /usr/lib/libkabinterfaces.so.1.0.0
> /user/lib/64/kde3/libkontact-kaddressbookplugin.so
> 
> after running fsck and rebooting, the problems above seemed
> disappeared.  Now, after powering up the pc again (KDE session,
> again):
> 
> X starts, kicker, kmail and konsole crash every time (I'm typing this
>  from mutt in an xterm which I started from the "run command" graphic
>  utility)
> 
> if I start kicker from the xterm I get:
> 
> [marco at pc ~]$ QGDict::hashAsciiKey: Invalid null key
> kicker: crashHandler called
> KCrash: Application 'kicker' crashing...
> QGDict::hashAsciiKey: Invalid null key
> 
> trying to start  kmail:
> 
> > kmail
> kmail: relocation error: /usr/lib64/libkleopatra.so.1: symbol _ZN4Kleo18KeySelectionDialogD1Ev, version GLIBCXX_3.4 not defined in file libstdc++.so.6 with link time reference
> 
> kate and konqueror, instead, start without problems (though I haven't
> done anything with them yet)
> 
> Unfortunately, I had no possibility since wednesday to stay at the
> computer more than a few minutes every time (if I had had more time
> and concentration I would have probably done things differently), and
> at this point, given the sequence of events (update partly failed -
> forced fsck - another update) I am not sure how to:
> 
> - find out _EVERYTHING_ wrong in the system right now (hanging lock or
>   tmp files, all corrupted binaries, which rpms should be reinstalled,
>   what else).
> 
> - what else to report to help you diagnose the problem
> 
> - how to fix it! If I were sure only one package is corrupted, I'd
>   remove it ignoring warnings and reinstall it right away, but this is
>   something more serious. Remove all packages updated since wednesday
>   and then run yum update again?!? No, unless I have confirmation it
>   makes sense and/or there is no other way
> 
> Thank you in advance for any feedback
----
no way that should have happened...you clearly have corrupted volume(s)
and you need to back up your documents (settings if desired) and clean
install.

The thing I would be worried about is why this happened because it would
seem likely to happen again.

In other words, I can't believe that an update would cause the
corruption unless the hard disk map was already broken and then the
updates merely exacerbated the problem.

Craig