Red Hat Bugzilla – Bug 510172
anaconda doesn't detect any properties of existing partitions
Last modified: 2009-07-21 19:21:11 EDT
Created attachment 350891 [details]
Description of problem:
Running 7 July 2009 boot.iso/install.img, anaconda does not detect existing filesystems and partition types already existing.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Create an installation using any Fedora version
2. Reboot and run the Rawhide install from boot.iso
Existing partitions are detected for the layout dialog, but nothing about them other than their existence.
The layout dialog should display all existing layout info.
My test installation had the following layout.
sda1 - /boot (ext2)
sda2 - swap
sda3 - lvm with / (ext2) and free space
Created attachment 350892 [details]
BTW, just in case it wasn't obvious, I meant the custom layout dialog in Comment 0.
Are you getting any kernel errors on tty4? I was seeing this same problem yesterday, but only on some installs - not all.
Created attachment 350973 [details]
syslog with HAL errors (floppy?)
I don't have any kernel errors on tty4.
I've attached the syslog from another installation attempt. The only things like an error are two HAL dumps that seem to be associated with the floppy.
Sure enough, I can't mount the floppy in tty2 no matter what options I feed mount. That seems like a different bug to me, though.
You've definitely got kernel errors in the log file, though, which is probably preventing HAL from giving us a complete view of hardware. Reassigning.
FYI, the kernel version is 2.6.31-0.42.rc2.fc12.i586 in the rawhide images.
So we've reached the point where we can't install on a machine with 256MB of memory? How is this the kernel's fault?
(In reply to comment #7)
> So we've reached the point where we can't install on a machine with 256MB of
> memory? How is this the kernel's fault?
The same result occurred with 512MB, although I don't have any log files from that trial.
The error is that anaconda doesn't detect any info about existing partitions (other than their number and size). I didn't try blowing away the existing layout, but there was every indication that such an installation would have succeeded. My interest was in reusing the existing partitions, however, so I stopped there.
To completely answer how this situation is "the kernel's fault" would require some telepathy on my part, but I believe CL's rationale is to eliminate kernel errors before attempting to debug anaconda.
I'm currently up against Bug 510966, so I cannot provide further info on this bug at this time.
This bug appears to be resolved in anaconda-12.3 and kernel-2.6.31-0.81.rc3.git4 with today's (21 July 2009) boot.iso.