Fedora 20 stopped booting after migrating to RAID array

asked 2014-04-12 17:39:59 -0600

mplucinski gravatar image

updated 2014-04-12 17:41:19 -0600


I have decided to migrate my working Fedora 20 installation onto RAID-1 software array. After this operation, I have a problem with booting my system.


My machine is MacBook Pro (Early 2011). I have two 2TB external disks where I decided to keep my linux. As boot manager, I am using rEFIind (http://www.rodsbooks.com/refind/), which automatically detects linux kernels and can boot them without using GRUB.

On the beginning, I installed Fedora 20 on disk #1 (sdc):

    /dev/sdc2 - LVM PV (/, /home, swap); 1.5TB
    /dev/sdc4 - /boot; 500GB

It has worked properly for a few weeks. However, I decided to mirror current installation with second disk. I made following steps:

  1. Partitioned disk #2 (sdd):

    /dev/sdd2 - linux raid, 500MB
    /dev/sdd3 - linux raid, 1.5TB
  2. Created two RAID devices:

    /dev/md0 - mirror: [missing], /dev/sdd2
    /dev/md1 - mirror: [missing], /dev/sdd3

    (I made them with one missing volume, as my idea was that after moving all data into raid, I can remove original partitions, and use their space as place for raid volumes)

  3. Copied directly (dd) /dev/sdc4 (/boot) into /dev/md0

  4. Moved data from /dev/sdc2 into /dev/md1 using LVM's pvmove
  5. Rebooted

I was prepared for troubles, but apparently fEFInd properly found kernels on both old and new partitions and was able to run them.

The problem

Unfortunately, system stops after few seconds and after a while informs that it was not possible to mount root filesystem. Doing cat /proc/mdstat shows that it does not find my raid devices:

    + cat /proc/mdstat
    Personalities : 
    unused devices: <none>

Apparently, partitions belonging to volumes seems to be properly detected

    /dev/sdd2: UUID="5b824683-4db3-8fc4-bfe7-8010bc810f04" TYPE="linux_raid_member" PARTUUID="a45057fd-c205-42d2-9e9d-f54de894ac04" 
    /dev/sdd3: UUID="6bba47f6-325c-8218-778b-5b2c16a638d1" UUID_SUB="d47589ff-88ab-5f0d-397a-63b030b61eb1" LABEL="localhost.localdomain:1" TYPE="linux_raid_member" PARTUUID="25506dc8-8e00-4105-873d-dba7b4725b59"

Here is rdsosreport.txt containing more information about my configuration and boot log: http://pastebin.com/EjaSwvp2

I suppose that system should automatically create /dev/md* following RAID metadata from detected volumes. However, it does not do it. Has anyone an idea, why it is happening?

edit retag flag offensive close merge delete