From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040312 Description of problem: Even after a kickstart install, firstboot runs system-config-display, because it doesn't find /etc/X11/XF86Config. Sure enough, this file is no longer created by the installer or by xorg-x11; it's now named xorg.conf. If I'm reading this code right, it will run system-config-display on *every* reboot, which certainly sucks. This is new in the 0421 compose; 0420's didn't have this problem. Version-Release number of selected component (if applicable): firstboot-1.3.11-1 How reproducible: Always Steps to Reproduce: 1.Do a fresh kickstart install of the 0421 compose 2.Wonder why system-config-display is run; tell it to keep the config 3.Reboot Actual Results: You get system-config-display again, I suppose. Expected Results: Shouldn't even have got it the first time. Additional info:
Ah, it was running system-config-display if /etc/X11/XF86Config does not exist. It should have been looking for xorg.conf instead. firstboot-1.3.12-1 in Rawhide should fix this.
I need to test and see if this affects non-kickstarted machines as well.
Yes, this affects non-kickstarted machines as well. Sopwith, I think that this should really be fixed before test3 goes out.
On second thought, it's just a minor annoyance. Let's get test3 out the door and I'll dupe the bug reports to this one.
It's actually a very minor annoyance, since there's no trace of firstboot in /etc/rc.d/rc*.d after the first boot. So my theory that it would happen on every boot was wrong; it's really only the first time.
*** Bug 122073 has been marked as a duplicate of this bug. ***
*** Bug 122041 has been marked as a duplicate of this bug. ***
*** Bug 121820 has been marked as a duplicate of this bug. ***
*** Bug 122422 has been marked as a duplicate of this bug. ***
This now happens with upgrades (from say RHL-9 or FC1), b/c the config is still named XF86Config. The real issue though is that firstboot is doing this even though it is supposed to be turned off. The initscript itself does not check /etc/sysconfig/firstboot; it just goes ahead and runs system-config-display.
Is firstboot ever getting run after doing the initial install? If so, firstboot should call "/sbin/chkconfig --del firstboot" when it completes which should make sure that the initscript never gets run again.
The scenario here is a kickstart install followed by a kickstart upgrade. There is never a user at the console, and firstboot should never get run. The problem (which I believe Elliot fixed) was that the firstboot initscript (not the actual python code) was running system-config-display even though /etc/sysconfig/firstboot indicated that firstboot was disabled.
Mike: which version of firstboot are you seeing this problem on? As far as I can tell, the change that Elliot made (contained in firstboot-1.3.14-1) should have fixed this problem. The initscript now contains: if [ ! -f /etc/X11/xorg.conf -a ! -f /etc/X11/XF86Config ] ; then echo -n $"X is not configured. Running system-config-display" /usr/bin/system-config-display which I believe will only run system-config-display only if both xorg.conf and XF86Config do not exist.
Yes, Elliot's changes did fix this bug. However, after a second look, it seems the changes may err a bit too far on the "don't run" side. Now the initscript will exit immediately if /etc/sysconfig/firstboot exists, even if reconfig is passed on the kernel cmdline. I'm attaching a patch that takes a more balanced approach.
Created attachment 101229 [details] a slightly more robust approach
Patch applied in firstboot-1.3.15-1. Thanks for the patch.