FC1 -> FC2 upgrade notes

Simon Andrews simon.andrews at bbsrc.ac.uk
Wed May 19 12:17:17 UTC 2004


I've just finished an upgrade from FC1 to FC2 on our development machine 
so I thought I'd post some notes about how it went.  Things are mostly 
OK, but there were one or two issues which came up during the upgrade.

Hardware is a Dell PE2500, Dual 1Gz Xeons running FC1 + all updates, 
plus some stuff from Livna.

The upgrade process was all smooth with no errors from the installer.

On rebooting for the first time kudzu prompted me to reconfigure my 
keyboard despite this being done during the install.

Upon logging in (via gdm) for the first time I got a load of popups saying:

"Error activating XKB configuration, probably an internal X-server problem"

This seems to be a manifestation of bugilla #120858, which was supposed 
to be fixed in the release, but doesn't seem to fix the upgrade of a FC1 
system, just a fresh install.

Interestingly, when I went to apply the suggeted fix I found that the 
file /etc/X11/xorg.conf didn't exist, and that X was still using 
XFree86.conf.  I changed the line which said:

"XkbRules"	"xfree86"

to be:

"XkbRules"	"xorg"

and restarted X, and everything was fine with that.

When I checked /var/log/messages I saw that during the first boot 
sendmail gave an error about timestamps being in the future.  I checked 
the time on the server and it seemed OK.  NTPd was running and seemed to 
be working, but restarting it produced a warning saying:

"keyword authenticate unknown, ignoring line"

I fixed this by commenting the line in /etc/ntp.conf which says:

authenticate yes

..which works for me as our ntp server doesn't require authentication, 
but others may need to find the new correct syntax.

I edited my /etc/yum.conf to point the updates at FC2 instead of FC1. 
Unfortunately the core and update directories for FC2 don't yet exist 
(at least not on the mirror we use), and yum dies when it gets a 404 
error, so for the time being I've had to comment out the core and update 
entries in yum.conf.  I imaging these directories will appear in the 
near future.

Also with yum, after I'd commented out the non-existant repositories I 
get a warning when running yum, saying:

"/etc/security/selinux/file_contexts : No such file or directory"

I imagine this is a hang over from when selinux was on by default.  It 
doesn't seem to stop anything from working though.

I had some stuff installed from livna (eg mplayer).  The Livna RPMs 
didn't seem to be taken into account when updating, meaning they were 
still installed (I was expecting them to have been removed after a 
warning), but they didn't work due to not having compatible dependencies 
installed.  I've removed these for now and will put them back when new 
versions appear.

Finally, the biggest breakage for us was that we run a load of CGI 
scripts which access a local mysql server.  All these have stopped 
working due to authentication errors.  The problem seems to be that 
although both the web server and the mysql server are on the same 
machine, apache is identifying itself using the address of eth0 rather 
than as localhost.  I've not been able to find an easy fix for this yet 
other than going through and replicating all the localhost permissions 
in the database to the external address of eth0.  Any suggestions of how 
to sort this out are welcome!

Generally though a pretty smooth transition.  Many thanks to all those 
involved in this release.





More information about the users mailing list