[Spacewalk-list] CentOS 5.2 - a warning]

[m.roth2006 at rcn.com]

>Date: Wed, 22 Apr 2009 21:26:44 +0100 (BST)
>From: John Hodrien <J.H.Hodrien at leeds.ac.uk>  
>On Wed, 22 Apr 2009, m.roth2006 at rcn.com wrote:
>
>> <snip>
>But you seem to be skipping my point and cutting the wrong bits from my post.
>
>The i686 glibc *should* be installed.  Pick a regular x86_64 install of
>CentOS/RHEL/Fedora and you'll find there's a glibc.i686 package installed.
>
>If you don't have a 32bit install of glibc installed then you'll not be
>running any 32bit binaries that rely on glibc, which is... everything apart
>from static binaries.
>
>Having the i686 glibc installed does not cause these problems.  It just
>doesn't.
>
>> And I did correct myself - there were some devel and other 686 kernel
>> packages, but no kernel-2.6...i686.rpm.
>
>If you didn't have a 32bit kernel package installed you wouldn't be seeing
>symptoms matching the link you provided.

Beg pardon? I'm running the 64-bit version of CentOS. Therefore, it has to be an incorrect 32-bit package that's installed, and what's being spoken to as the system comes up. This would make sense, if a 32-bit variable was being handed a 64-bit value.

I'll repeat myself: no matter what I did, the system *hung* on boot, until I took out the i686 package, and then there were zero failures on the upgrade, and the boot had no problem. Were you correct, it would not have failed in the first place. What do you suggest that would lead me to any other conclusion? 

       mark