Description of problem:
Root partition on dmraid RAID1 device (pdc_bcdbecjef). 
During early boot (dracut) the underlying partitions (sda?, sdb?) should be removed by dmraid --rm_partitions ... This fails however. With the sdaX and sdbX devices still present, udev will start lvm2-pvscan@.service on these devices. It seems that this actually leads to the usage of sdaX or sdbX for the LVM physical volume instead of pdc_bcdbecjefX


Version-Release number of selected component (if applicable):
dmraid-1.0.0.rc16-25.fc21.x86_64
lvm2-2.02.115-1.fc21.x86_64

How reproducible:
always

Steps to Reproduce:
1. boot system with logical volume on dmraid (e.g., /home)
2.
3.

Actual results:
LVM's physical volume is /dev/sdaX

Expected results:
LVM's physical volume is /dev/pdc_bcdbecjefX


Additional info:

If everything works fine dmsetup ls --tree shows (on my system):
pdc_bcdbecjef1 (253:1)
 └─pdc_bcdbecjef (253:0)
    ├─ (8:16)
    └─ (8:0)
vg02-system (253:5)
 └─pdc_bcdbecjef4 (253:2)
    └─pdc_bcdbecjef (253:0)
       ├─ (8:16)
       └─ (8:0)
vg02-home (253:6)
 └─pdc_bcdbecjef4 (253:2)
    └─pdc_bcdbecjef (253:0)
       ├─ (8:16)
       └─ (8:0)

'/' on /dev/mapper/pdc_bcdbecjef1
'/home' on /dev/mapper/vg02-home

I disabled mounting '/home' in /etc/fstab. Then I get after boot:

> systemctl --all | grep pvscan 

 lvm2-pvscan@253:2.service      loaded    active   exited    LVM2 PV scan on device 253:2  
  lvm2-pvscan@8:20.service       loaded    active   exited    LVM2 PV scan on device 8:20   
● lvm2-pvscan@8:4.service        loaded    active  activating  LVM2 PV scan on device 8:4
  system-lvm2\x2dpvscan.slice    loaded    active   active    system-lvm2\x2dpvscan.slice 

> pvdisplay
  --- Physical volume ---
  PV Name               /dev/sda4
  VG Name               vg02
  PV Size               830,82 GiB / not usable 1,31 MiB
  Allocatable           NO
  PE Size               4,00 MiB
  Total PE              212690
  Free PE               0
  Allocated PE          212690
  PV UUID               pRAI6d-g1LF-Tqrs-sYVd-no1X-XF4f-PwCIQY

To repaired this, I did the following:
> systemctl stop lvm2-pvscan@8:4.service
> vgchange -an vg02
> pvchange -x n /dev/sda2
> pvscan --cache -a ay 253:2

With this:
> pvdisplay
--- Physical volume ---
  PV Name               /dev/mapper/pdc_bcdbecjefpdc_bcdbecjef4
  VG Name               vg02
  PV Size               830,82 GiB / not usable 1,31 MiB
  Allocatable           NO
  PE Size               4,00 MiB
  Total PE              212690
  Free PE               0
  Allocated PE          212690
  PV UUID               pRAI6d-g1LF-Tqrs-sYVd-no1X-XF4f-PwCIQY

So the dmraid array would be working, if it would be activated properly!

Being astonsished that sda4 (and sdb4) is still present, I rebooted with 'rd.break=initscripts'. Until this point /dev/mapper/pdc_bcdbecjef was not yet present, as well as no /dev/sda4 or /dev/sdb4. Analogue to dracut's dmraid-module I performed:

dmraid -ay -i -p --rm_partitions pdc_bcdbecjef

Unexpectedly, /dev/sda1, /dev/sda4, /dev/sdb1 & /dev/sdb4 were created by this command!!