If a scriptlet error occurs during an upgrade, a dialog appears saying: msg = _("An error occurred while installing packages. Please " "examine /root/install.log on your installed system for " "detailed information.") But that's the wrong logfile name during upgrade. You probable should use instLog. Here's where that code was added: http://www.redhat.com/archives/anaconda-devel-list/2010-September/msg00163.html
This may not be perfect, but does the job. --- yuminstall.py.orig 2011-04-22 21:44:38.000000000 -0400 +++ yuminstall.py 2011-04-22 21:46:52.000000000 -0400 @@ -783,8 +783,8 @@ else: if rc.return_code == 1: msg = _("An error occurred while installing packages. Please " - "examine /root/install.log on your installed system for " - "detailed information.") + "examine %s on your installed system for " + "detailed information.") % instLog.name.replace('/mnt/sysimage','',1) log.error(msg) if not self.anaconda.isKickstart: You'd need to change all the L18N as well. [Now I just need to find out why CentOS 5.6's kod-gfs-PAE i686 rpm is generating this exception on every upgrade - with nothing logged in the upgrade log.]
This is fixed in the next major release of RHEL (namely, RHEL6). If you require a fix to RHEL5, please raise this issue through your support representative who will in turn make sure it's on our schedule. Thanks.