Mcguffey, David C wrote: > On 12 Oct 2005 I reported in a message (Subject: FC4 post-yum message > "Kernel panic -- not syncing: Attempted to k ill init!") that yum was > crashing my FC4 load to the point it wouldn't boot. > > Today, I finally tracked down the fact that when yum retrieves a kernel > update (2.6.11 to 2.6.13) for i686, the kernel won't survive a reboot. It > will continue to run after yum is finished, but a reboot causes the message > reported above. Can this be some kind of problem generating or using the kernel initrd? What can be seen on screen in the few lines before this error? Is there room in /boot? > First question is related to the i686 verses i386 designations. All of the i386 is the lowest common denominator which works on i586 and i686 machines too, most packages see no benefit in making differently compiled versions for the more advanced CPU types, so most packages are compiled for i386 even if the host CPU is, say, an i686. The kernel does benefit from being compiled for i686 if you have an i686-class CPU. If the problem was that you had an i386 or i586 -class CPU and were using the i686-compiled kernel, it would blow chunks in a more spectacular fashion long before this panic. > Last question (I know I shouldn't ask more than one question in a post), > what is the proper yum command line to exclude the two kernel updates? I > want to yum everything but the kernel. just yum --exclude kernel update should do it if I understand you correctly. However it sounds either like the problem is a munged initrd generation (the /boot/initrd* file for the new kernel) or a problem to do with getting the right storage device driver into the initrd, and not related to whether yum updated the kernel package on the previous boot directly. Why does yum keep trying to insert a newer kernel? sounds like the kernel install failed... -Andy
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
