Description of problem: AFAICS liveinst rewrites ifcfg-eth0 based on the current values from the running livecd. That is generally very nice. BUT it leaves out NM_CONTROLLED. That is unexpected and problematic in some cases when creating installable custom livecds (which I was doing). Version-Release number of selected component (if applicable): anaconda-11.4.1.63-1.i386
I notice some awareness of NM_CONTROLLED in /usr/lib/anaconda/network.py, but it seems like it doesn't try to write NM_CONTROLLED. Probably unrelated, but AFAIK _only_ NM_CONTROLLED=no (case sensitive) has any semantics to NM, so the checks for yes case insensitive or empty is not correct.
Created attachment 346166 [details] 0003-Test-NM_CONTROLLED-setting-correctly-in-network.py.patch Patch to preserve NM_CONTROLLED variable on custom live images. Awaiting patch review.
Patched committed to master branch in git, will be in the next anaconda build for rawhide.
Thanks. I noticed the actual NM parsing is done in http://cgit.freedesktop.org/NetworkManager/NetworkManager/tree/system-settings/plugins/ifcfg-rh/reader.c#n2105 . Actually both n, no and false can be used. It would be simpler if only one was supported, but perhaps you want to support all of them.