With anaconda 17.19, which is intended to address outstanding showstoppers http://bugzilla.redhat.com/show_bug.cgi?id=810005 and http://bugzilla.redhat.com/show_bug.cgi?id=810391 in preupgrade, a new issue arises: again boot to anaconda fails during dracut stage because initqueue fails. This bug has been discussed in http://bugzilla.redhat.com/show_bug.cgi?id=810391 comments #5 to #11. The issue appears to be that the kickstart file loading fails. Marking as a Beta blocker, we have solid history for accepting showstopping preupgrade issues as beta blockers.
Having both "stage2=hd:UUID=XXX" and "ks=hd:UUID=XXX" meant we had two jobs to perform when the disk with UUID=XXX appeared. Unfortunately, initqueue did unexpected things with the job names such that the two jobs collided. Try this and tell me if it fixes the problem: - boot with 'rd.break=cmdline' - when the shell appears, do: sed -i 's/--unique//' /lib/anaconda-lib.sh exit The patch I'm proposing does, basically, that. If you can confirm that works, we can skip the process of filing a new blocker bug when the next unexpected bug appears. :D
Created attachment 576325 [details] init.log after applying the 'workaround' suggested by wwoods I tried wwoods' suggested workaround without success: 09-04-2012 14:34:01 < wwoods!wwoods@nat/redhat/x-sxoulxstnhtqfwjd: adamw: if you want to test the fix, it's pretty easy: boot with rd.break=cmdline, then: 09-04-2012 14:34:21 < wwoods!wwoods@nat/redhat/x-sxoulxstnhtqfwjd: sed -i 's/--u nique//' /lib/anaconda-lib.sh 09-04-2012 14:34:37 < wwoods!wwoods@nat/redhat/x-sxoulxstnhtqfwjd: then exit the shell and it should start the installer still get an initqueue failure, with a devexists-\\x2fdev\x2froot.sh that fails and also a kickstart.sh that still fails. There is no /tmp/ks.cfg or /tmp/ks.cfg.done. Here is the init.log.
Created attachment 576339 [details] test run with anaconda 17.19 and edited anaconda-lib.sh I added rd.break=cmdline rd.debug console=ttyS0 to the cmdline (rd.debug doesn't seem to log properly when used with rd.break=cmdline). I edited anaconda-lib.sh to remove the --unique flag and to add the readlink modification to dev_is_mounted.
Created attachment 576354 [details] additional debugging added rd.udev.info and rd.udev.debug Things seem to go wrong around line 11248 where the initqueue being run from udev.
-- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
It turns out that also doing: sed -i 's/1-.dev/1/' /lib/anaconda-lib.sh to replicate the other part of wwoods' patch results in success. He didn't think it'd be important, but it is.
anaconda-17.20-1.fc17 has been submitted as an update for Fedora 17. https://admin.fedoraproject.org/updates/anaconda-17.20-1.fc17
Package anaconda-17.20-1.fc17: * should fix your issue, * was pushed to the Fedora 17 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing anaconda-17.20-1.fc17' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/FEDORA-2012-5559/anaconda-17.20-1.fc17 then log in and leave karma (feedback).
anaconda-17.20-1.fc17 has been pushed to the Fedora 17 stable repository. If problems still persist, please make note of it in this bug report.