Bugzilla – Bug 147415
mkinitrd fails during installation if root filesystem is on LVM
Last modified: 2006-02-02 07:37:14 UTC
root filesystem on lvm1 mkinitrd thinks it hit a lvm2 root volume and writes the wrong linuxrc. Even if it were a lvm2 volume, there is still an error in line 2316: |if [ -z "\$lv_root" ]; then this should be |if [ -z "\$vg_root" ]; then because $lv_root will always be empty because the lines above that set $vg_root.
sorry, should have assigned it to Hannes directly.
SGI has hit this with ia64 on Altix, we believe. In the ia64 case, the boot files are missing from /boot/efi even though /boot/efi isn't part of the volume of course.
[I asked our QA team if their bug could be the same as this already filed suselinux one, Jim Grisham in SGI QA replied:] Could be. I wasn't able to reproduce the hang after deleting/creating a new lvm. It installed now and I verified /boot/efi has no files in it. Here's the info from partition in the installer: Delete logical volume /dev/system/system (58.8 GB) Remove volume group system Delete partition /dev/sde3 (24.6 GB) Delete partition /dev/sdf1 (34.1 GB) Create partition /dev/sde3 (24.6 GB) with id=8E Create partition /dev/sdf1 (34.1 GB) with id=8E Create volume group lvmroot Create logical volume /dev/lvmroot/lvmroot (58.8 GB) for / with reiserfs Format partition /dev/sde1 (517.6 MB) for /boot/efi with vfat Use /dev/sda10 as swap Use /dev/sdb7 as swap Use /dev/sdc2 as swap Use /dev/sde2 as swap Use /dev/sdg7 as swap Use /dev/sdh7 as swap Use /dev/sdi7 as swap Use /dev/sdj7 as swap Here's what was mounted during install, includes /dev/sde1: inst-sys:~ # mount rootfs on / type rootfs (rw) tmpfs on / type tmpfs (rw) tmpfs on / type tmpfs (rw) proc on /proc type proc (rw) sysfs on /sys type sysfs (rw) /dev/loop0 on /mounts/instsys type cramfs (ro) devpts on /dev/pts type devpts (rw) none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw) 163.154.12.145:/install/sles10 on /var/adm/YaST/InstSrcManager/IS_CACHE_0x00000001/MEDIA type nfs (ro,nolock,addr=163.154.12.145) /dev/sde1 on /mnt/boot/efi type vfat (rw) /dev/mapper/lvmroot-lvmroot on /mnt type reiserfs (rw) proc on /mnt/proc type proc (rw) sysfs on /mnt/sys type sysfs (rw) Stopping it before the first reboot, mount shows: nst-sys:~ # mount rootfs on / type rootfs (rw) tmpfs on / type tmpfs (rw) tmpfs on / type tmpfs (rw) proc on /proc type proc (rw) sysfs on /sys type sysfs (rw) /dev/loop0 on /mounts/instsys type cramfs (ro) devpts on /dev/pts type devpts (rw) none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw) /dev/mapper/lvmroot-lvmroot on /mnt type reiserfs (rw) sysfs on /mnt/sys type sysfs (rw) ...Remounted /dev/sde1 and no files: mount /dev/sde1 /mnt/boot/efi inst-sys:~ # cd /mnt/boot/efi inst-sys:/mnt/boot/efi # ls inst-sys:/mnt/boot/efi # Thanks for tracking that down! [snip]
No, your bug is bug 147366 "Installer mounts filesystems in wrong order, resulting in boot failure". Quoting from your report: inst-sys:~ # mount [...] /dev/sde1 on /mnt/boot/efi type vfat (rw) /dev/mapper/lvmroot-lvmroot on /mnt type reiserfs (rw) [...] That's the wrong mount order bug. However, you would have been hit by this bug if you had solved the other one.
Created attachment 66141 [details] mkinitrd Fixed /sbin/mkinitrd
Fixed mkinitrd submitted to STABLE for SLES10 Beta4. In the meantime you can use the attachment.