Hi all, I just spend (lost) way too much time on some grub weirdness, so I thought it would be a good idea to write it down somewhere. I did a test install on a system with a 80 GB fakeraid set. The fakeraid set had: p1: 200 MB ext3 /boot p2: 79GB PV I removed the PV and created (slightly simplified): p1: as it was p2: 74GB ext3 / (also /boot) p3: 200 MB swap p4: 2200 MB swap First it took me some time to figure out I was still getting the grub from p1 (which initrd could no longer find /), somehow the setup of the new grub (which lives on p2) failed. Then I tried to setup grub manually (using a device.map for the dmraid BIOS mapping), but when I entered: grub> root (hd0,1) grub> setup (hd0) Grub complained it could fined neither of /grub/stage1 and /boot/grub/stage1, where as ls clearly shows /boot/grub/stage1, it can be successfully cat-ed etc. Which explains why the setup failed, but why does grub not find stage1? So now I tried to use p3, the 200 MB swap as /boot, grub now complained it used cylinders beyond the BIOS address limit. It seems the BIOS cannot address cylinders above circa 10240, do we sanity check our /boot (or /) for this ? So I reformatted p1, copied all the stuff from p2 /boot there and setup grub (that finally worked). Then when I rebooted things worked, after some minutes I realized I had not updated grub.conf, so grub as run from the BIOS was still referring to p2 for vmlinuz and the initrd, and it worked. So: grub run under Linux (the grub shell) cannot find stage1, but run under the BIOS (so stage2) can find vmlinuz and initrd ???? Regards, Hans _______________________________________________ Anaconda-devel-list mailing list Anaconda-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/anaconda-devel-list
