Fedora
RPMs
rpmdb: fatal region error detected; run recovery
Short version: "rm -f /var/lib/rpm/__db*".
Longer version: Remove stale lock files.
Link: http://www.internetguru.com.au/igblog-127.html
Problem Kernels
2339
Can't boot Zephir or Edison1. Edison1 boots OK if using kexec some of the time
Fedora 6 Upgrade
Dog slow and wierd with CDs.
Fine over network.
Using Updated Install Source
Instead of installing the relese version of everything and then getting new versions later with yum, I like to create an install source that uses the latest RPMS, so my new install are up-to-date immediatly. With previous versions of Fedora (and therefore Anaconda), I had to winnow out old versions of all the RPMs and also run a cumbersome set of scripts to create the files that make the collections of RPMS into a valid install source. Not with Fedora 6!
I first made a mirror of core and updates for FC6. Then I copied core to another directory, using cp -l. The '-l' switch tells cp to create hard links rather than actually copying the file, saving disk space. Then I used cp -l again to copy the RPMS from updates to my new directory's RPMS subdir. Then I used createrepo on my new directory to turn it into a working Yum repo.
I just performed an install against this new blended repo, and it worked great. I haven't tried using as an upgrade source yet, though.
Removing Network Interfaces
Just removing /etc/system-config/network-scripts/ifcfg-foo doesn't seem to work any more. Apparently, there's /etc/system-config/profiles and /etc/system-config/devices dirs containing redundant and confusing ifcfg files. A device must be deleted from these as well...