On Tue, Nov 17, 2015 at 11:14 AM, Felix Miata <mrmazda@earthlink.net> wrote:
CPU is Intel Haswell on B85 chipset. Multiboot RAID1 installation started with 13.1 on /dev/md1, 11.4 on /dev/md2, and 13.2 on /dev/md3.
Wishing to test a DVD media upgrade from 32 bit to 64, booted to 13.1, I umounted 11.4 (on EXT3), reformatted md2 EXT4, remounted md2, then rsync'd from 13.2 on md3 to the freshly formatted md2. Next I adjusted what was the 11.4 grub stanza on /dev/sda1 to do the new 13.2 on md2, and fstab on md2.
When I tried to boot md2, I always got a Dracut emergency shell reporting [/dev/md2|LABEL=10os13264] root device does not exist. After a few repeats, I booted the original 13.2 on md3, chrooted into the new 13.2 on md2, changed dracut's configs to hostonly="no" and hostonly_cmdline="no", ran dracut, and tried booting 13.2 on md2 again, with identical result, except taking 6+ seconds instead of 4+ seconds to reach the fail point due to the monstrously larger non-hostonly initrd.
The dracut shell seems to be produce a runaround trying to mount anything to copy rdsosreport.txt to. It has no fdisk to report anything about partitioning. /dev/md* does not exist.
'systemctl reboot' apparently won't complete, as it's trying to undo things that it failed to do in the first place. CAD doesn't do anything. 'exit' doesn't cause reboot either.
Finally I tried another chroot initrd build, this time with hostonly='no', and with add_drivers+="raid1 md_mod". It works, but I don't understand why the explicit addition of the two extra modules to the initrd was required, or
One reason could be that your chroot did not have /dev, /sys and /proc required for autodetection of kernel drivers.
why the failure with the comprehensive hostonly=no initrd. If this kind of clone can't be booted without all the fuss, how could a restored backup? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org