I just installed UBUNTU (I am new to UBUNTU & linux) on a 40GB SATA - sdd, I have additional 3 SATA drives 1.5 TB each = sda, sdb,sdc and I want to set LVM over RAID 5, I intend to assign the whole drive for the RAID.
1. Do I need to partition the drives for setting the RAID or I can go ahead and set the RAID5 over the whole drive devices without setting any partitions, from my search I understand that both is possible but not clear on which is better and why?
2. Assuming the above is that I need or should partition. Which command is the best for using script to do the partitioning: "parted" or "sfdisk" (or other command...)
3. What should be the partition type FD or DA or other type?
4. If I am using "parted" command to do the partitioning how do I set the type to FD or DA etc using parted or I must use sfdisk for doing RAID and LVM?
5. I understand that sfdisk can't handle large drives, can I still use it and set on each of my drives multiple / smaller partitions each can be created by sfdisk and create multiple RAID5 arrays based on the partitions and than assign all the RAIDs to the LVM so the LVM still see the whole storage? Does this extra partitioning and RAIDs affect performance?
6. What performance can I expect from LVM over RAID 5 if my system is ASUS E35M1 motherboard with AMD E350 Dual Core CPU and 4GB RAM SATA 3G drives for the RAID/LVM and SATA 1.5G for the system drive?

Do wait for other people's comments, but

1. best practice is to create a partition across each disk

2. don't know about 'best', personally I've always used fdisk.; YMMV

3. RAID uses type fd

Note:
generally, 3 drives is min for RAID5, but this would mean that when a disk goes bad you'll need a backup system.
You may(!) be able to recover from 2, but I wouldn't bet on it.
You can add an extra drive as hot standby.

https://en.wikipedia.org/wiki/Standard_RAID_levels
https://en.wikipedia.org/wiki/RAID

HTH

RAID5 always has one drive of redundancy. It doesn't matter how many drives you have in your array, you can always lose one without losing any data and rebuild from the rest of them with RAID5. If the entire array crashed after losing a single drive, that would be RAID0 (aka zero redundancy, an accident waiting to happen).

Chris,
Thanks for the details.

The reason for my question about FD or DA type is the following comment I found on other thread-

"(From the mdadm 2.6.8 man-page) When creating partition-based arrays using mdadm and version-1.x superblocks, the partition type should be set to 0xDA (non fs-data). This type selection allows for greater precision since using any other type [RAID auto-detect (0xFD) or a GNU/Linux partition (0x83)], might create problems in the event of array recovery through a live cdrom."

any idea about this array recovery problem when using FD type, I guess what is being suggested is that during the recovery process the system trys to find file system on each of the RAID drives if they are set as FD while if they are set as DA it does not and manage to move forward?

Thanks

Dov

First time I've heard of that; all the references say DA = 'non-FS data' & FD = linux raid auto.
I guess I'd check that you do have that version of mdadm & 1.x superblocks...

Let us know how that goes

It does not define any version for the mdadm and regarding to the superblock, I have no idea what is the version, actually I don't have any version of superblock yet since I din't start creating the partitions...
It is a lengthy process, I prefare to do it right with the help of people like you that have experience with UBUNTU/linux.

Any Idea what performance can I expect from software RAID 5 on my system? or otherwise on systems that you had experience with?

Thanks
Dov