Description of problem: Version-Release number of selected component (if applicable): 6.0.83-1 How reproducible: Create a initrd img and boot. Steps to Reproduce: 1. Install rawhide or create an initrd.img 2. Boot kernel. 3. Actual results: Boot hangs at Creating initial device nodes Expected results: Boot works. Additional info: Even though the boot hangs the system is minimally functional. I can do sysrq requests and ctrl-alt-del to reboot but that is all (no init was run I guess). This is an x86_64 system. I installed mkinitrd 6.0.71-4 from F10 and the system boots fine.
Can you please, update mkinitrd back to the latest again and then do: mkinitrd -v /boot/test.img $(uname -r) &> log And attached the generated log file here? Thanks! Can you also please attach your /etc/fstab, and tell us what kind of hardware your rood file system lives on (sata / ata / dmraid ... ?)
Okay, I did that and I noticed that running mkinitrd from my booted system generates a correct initrd image while using the rescue/install environment does this incorrectly. THis might be the case since the mkinitrd logs from the rescue environment are missing a root/swap disk reference. The root is on a sata drive (ahci).
Created attachment 342564 [details] fstab
Created attachment 342565 [details] booted mkinitrd log
Created attachment 342566 [details] rescue disk mkinitrd
Ah, yes this is due to the rescue (and install) environment missing a populated /dev, which is fixed by this commit: http://git.fedorahosted.org/git/?p=anaconda.git;a=commitdiff;h=57352b956a84d723c423e732efad8f9311c0d830 Which is in anaconda-11.5.0.49, whichc should show up in rawhide rsn, closing.
Hmm, I tried an install with this morning's rawhide images which indicate that they contain 11.5.0.49 yet I still have this problem (or something very much like it). It is trivial for me to bring up a guest install and reproduce this, but it seems pretty tough to debug it because I can't interact with the freshly installed system except through grub and there doesn't seem to be any way to coax the initrd to provide more info about what it's doing.
No luck with today's images that have anaconda 11.5.0.50, either. The behavior is unchanged. I wonder if I should file a separate bug.
(In reply to comment #8) > No luck with today's images that have anaconda 11.5.0.50, either. The behavior > is unchanged. I wonder if I should file a separate bug. Yes please file a separate bug, also please do an "ls /mnt/sysimage/dev" from tty2 while the installer is installing packages. Also while I'm asking for info, please boot into rescue mode, then make sure you've got a populated /mnt/sysimage/dev, and if its properly populated, chroot into /mnt/sysimage and run mkinitrd -v test.img $(uname -r) &> log there, and then attach the log file to the new bug report. If /mnt/sysimage/dev is not properly populated in either case, this is not an mkinitrd bug, but an anaconda bug.