Hide Forgot
Description of problem: After running preupgrade to Fedora 15, and rebooting into anaconda, it says that there is a problem with my LVM data, and if I want to re-initialize all my disks. After ignoring this, it gives the error of "the root for the previous installed system was not found" Version-Release number of selected component (if applicable): anaconda (shipped with fedora 15) fedora-release-14-1.noarch preupgrade-1.1.9-1.fc14.noarch How reproducible: It may depend on the LVM configuration Steps to Reproduce: 1. Install fedora 14 with a complex LVM configuration (1 volume group, several PV's (1 PV is an MD device) 2. run preupgrade 3. reboot to anaconda Actual results: upgrade fails Expected results: upgrade working Additional info: I'm attaching all the anaconda log's
Created attachment 501859 [details] storage log
Created attachment 501860 [details] anaconda log
Created attachment 501861 [details] program log
Created attachment 501862 [details] anaconda syslog
Created attachment 501863 [details] fdisk -l
Created attachment 501879 [details] lvs -v output
Created attachment 501880 [details] df -h
looking at storage.log, anaconda has a problem with my physical volume under a raid device. 18:27:17,044 DEBUG storage: lvm filter: adding md0p1 to the reject list 18:27:17,045 DEBUG storage: ignoring partition md0p1 18:27:17,116 WARN storage: Skipping LVs for incomplete VG vg_mirror 18:27:21,372 DEBUG storage: PartitionDevice.removeChild: kids: 1 ; name: sda2 ; 18:27:21,374 DEBUG storage: PartitionDevice.removeChild: kids: 1 ; name: sdc2 ; 18:27:21,376 DEBUG storage: PartitionDevice.removeChild: kids: 1 ; name: sdd1 ; 18:27:21,376 DEBUG storage: lvm filter: adding vg_mirror to the reject list 18:27:21,377 DEBUG storage: lvm filter: adding sda2 to the reject list 18:27:21,378 DEBUG storage: lvm filter: adding sdc2 to the reject list 18:27:21,378 DEBUG storage: lvm filter: adding sdd1 to the reject list
As a workaround, I backed up the content of my LV's hosted in the md0p1 PV. I removed md0p1 from the same volume group I had the root mount point. I created a new VG, and restored all my data there. upgrade was done smoothly, without any warnings.
Partitioned md arrays are not supported at this time except for certain varieties of bios/firmware raid which use md instead dmraid. I'm glad you were able to work around the issue and complete your upgrade. There is an existing RFE for support for partitioned md in bug 507578. *** This bug has been marked as a duplicate of bug 507578 ***