how to install from a raid5 strorge disk

[Farkas Levente]

hi,
we'd like reinstall some of our centos-6 servers remotely. setup:
- one system disk (sda) with a few partition (boot, swap, root)
- storage disks running one big software raid5 or raid6 partition.

what we'd like to do copy:
1. copy centos-7 installer: vmlinuz and initrd.img from
7/os/x86_64/images/pxeboot/vmlinuz
7/os/x86_64/images/pxeboot/initrd.img
7/os/x86_64/LiveOS/squashfs.img
to the system disk's root partition

2. copy the full installation tree with the repo to the storage
partition and use it as a source during install.


3. create a new grub entry with this options:
------------------------
title Install Centos7 KICKSTART MINIMAL
root (hd0,1)
kernel /ds/vmlinuz inst.stage2=hd:LABEL=root:/ds/
ks=hd:LABEL=root:/ds/minimal.ks
initrd /ds/initrd.img
------------------------
or something like this.

4. in the minimal.ks's %pre section mount the storage's raid6 filesystem
into /mnt/repo and in the same file use it as installation source ie:
file:///mnt/repo/...

it'd have to work in theory BUT the problems:

- if we DO NOT specify inst.stage2 then the system can't find stage2
even if we specify inst.repo=hd:... so it seem if the installation is
not network based (ie. nfs, ftp, http) the we must specify inst.stage2.
is it true?

- if we DO specify inst.stage2 then anaconda loop mount it and therefore
we can't put the stage2 file into the system disk's any partition since
after this anaconda can't umount the given partition so can't
repartition the system disk.
is there any was to force that anaconda use the memory (ie ramfs) even
in case of none network install ie: inst.repo=hd:..?

- is it possible to put stage2 into the raid6 storage partition?

- is there any other way to solve this problem?

thank you for your help in advance.
regards.