Description of problem: Upgrading a recent fresh install of FC6 to F7/development (buildstamp 200702220434.i386) fails to detect boot loader (grub) installed by FC6. Version-Release number of selected component (if applicable): 11.2.0.26 How reproducible: didn't try Steps to Reproduce: 1. boot anaconda installer 2. select upgrade of FC6 partition 3. Actual results: "The installer is unable to detect the boot loader currently in use on your system. What would you like to do? (greyed out: Update boot loader configuration) Skip boot loader updating Create new boot loader configuration Expected results: Anaconda should detect my existing grub in /dev/sda (which was /dev/hda on the FC6 kernel). Additional info: HP OmniBook 6000, P-III 600 MHz, 384 MB RAM.
Created attachment 148626 [details] First 64 sectors of /dev/sda
Output from sfdisk -x -l: Disk /dev/sda: 4864 cylinders, 255 heads, 63 sectors/track Warning: extended partition does not start at a cylinder boundary. DOS and Linux will interpret the contents differently. Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0 Device Boot Start End #cyls #blocks Id System /dev/sda1 0+ 1- 2- 15088+ a0 IBM Thinkpad hibernation end: (c,h,s) expected (1,224,63) found (29,15,63) /dev/sda2 * 1+ 2432- 2432- 19527480 c W95 FAT32 (LBA) start: (c,h,s) expected (1,225,1) found (30,0,1) end: (c,h,s) expected (1023,254,63) found (1023,15,63) /dev/sda3 2432+ 2446- 14- 105840 83 Linux start: (c,h,s) expected (1023,254,63) found (1023,239,63) end: (c,h,s) expected (1023,254,63) found (1023,239,63) /dev/sda4 2446+ 4863 2418- 19421640 5 Extended start: (c,h,s) expected (1023,254,63) found (1023,239,63) end: (c,h,s) expected (1023,254,63) found (1023,239,63) /dev/sda5 2446+ 4863 2418- 19421608+ 8e Linux LVM start: (c,h,s) expected (1023,254,63) found (1023,239,63) end: (c,h,s) expected (1023,254,63) found (1023,239,63) - 2446+ 2446- 0 0 0 Empty - 2446+ 2446- 0 0 0 Empty - 2446+ 2446- 0 0 0 Empty
Hmm looking at this partition table further, it seems that under FC6, the disk is detected with 16 heads: Disk /dev/hda: 77520 cylinders, 16 heads, 63 sectors/track But under FC7, the disk is detected with 255 heads: Disk /dev/sda: 4864 cylinders, 255 heads, 63 sectors/track Additionally, I believe some partitions were somehow created with 16 heads, and others with 240 heads in the partition table. This smells like an old anaconda/parted bug from ages ago.
... let me guess, /etc/sysconfig/grub has "boot=/dev/hda" in it.
Yes it does. How will upgrades handle this case?
Right now, not very well. We definitely need to do something about this for F7, just not quite sure what yet.
*** Bug 234834 has been marked as a duplicate of this bug. ***
*** Bug 235327 has been marked as a duplicate of this bug. ***
*** Bug 236476 has been marked as a duplicate of this bug. ***
We've committed a lot of stuff today to help handle these cases. Testing is still needed, though.
I tested today's (2004-04-19 10:00) rawhide install and I still get the same error upgrading a freshly installed, fully updated FC6: "The installer is unable to detect the boot loader currently in use on your system." anaconda-11.2.0.52-1 Here are my FC6 files: /etc/sysconfig/grub: boot=/dev/hda forcelba=0 /etc/grub.conf: # grub.conf generated by anaconda # # Note that you do not have to rerun grub after making changes to this file # NOTICE: You have a /boot partition. This means that # all kernel and initrd paths are relative to /boot/, eg. # root (hd0,2) # kernel /vmlinuz-version ro root=/dev/q/root # initrd /initrd-version.img #boot=/dev/hda default=0 timeout=5 splashimage=(hd0,2)/grub/splash.xpm.gz hiddenmenu title Fedora Core (2.6.20-1.2944.fc6) root (hd0,2) kernel /vmlinuz-2.6.20-1.2944.fc6 ro root=/dev/q/root rhgb quiet initrd /initrd-2.6.20-1.2944.fc6.img title Fedora Core (2.6.18-1.2798.fc6) root (hd0,2) kernel /vmlinuz-2.6.18-1.2798.fc6 ro root=/dev/q/root rhgb quiet initrd /initrd-2.6.18-1.2798.fc6.img title Fedora Core Development Installer root (hd0,2) kernel /vmlinuz-devel graphical askmethod initrd /initrd-devel.img title Other rootnoverify (hd0,1) chainloader +1
Created attachment 153054 [details] anaconda.log
Created attachment 153055 [details] anaconda syslog
The message is correct and we pretty much have to punt on doing automatic bootloader reconfig. While it's doable for the case of one controller, the second you throw a second controller (or even two disks) into the picture, it becomes impossible. So we detect and require you to generate a new boot loader config at which point we can get things correct. With today's rawhide, there's one little thing that's not right which is /boot/grub/device.map doesn't get properly recreated. But that should be fixed for tomorrow
Testing I've done on this is looking good. Closing out
I am seeing this situation "cant detect your bootloader" with f7t4 pxeboot+ftp on an LSI raid dell poweredge. Did the fixes in comment 14 make it in to test4 ?