SuSE 11.1 & Hardware Raid 0

markyb · 28 Sep 2009 at 15:51

Hey,

I'm in the process of building up a fileserver for production using SuSE 11.1.

There are 2x 160Gb SATA drivers configured in RAID1 (MCP55?) The machine is a HP MP115 G1.

I setup the RAID via the Nvidia setup before the machine booted (f8) and SuSE saw the hardware raid and installed without any issues. Its running well but a little concerned there is no indication of any of the drives/RAID failing.

Using Windows, Nvidia media sheild used to inform you that one of the drives were off-line or the raid was de-graded. After testing I only get this message if I reboot which is not practical.

Any advice?

BigglesPiP · 28 Sep 2009 at 15:56

That's not hardware RAID, it's hybrid RAID.

The way it is supposed to work is that the "RAID controller" presents a RAID device, and the computation is all done by your CPU via a Driver.

I suspect SuSE, in the lack of a Linux driver, will be using dmRAID, which accesses the individual drives and performs software RAID on them.

Might as well just use Linux Software RAID, or LVM.

markyb · 28 Sep 2009 at 16:04

BigglesPiP, Thanks for the reply.

- Would it be better for me to start again but this time using a software RAID using SuSE?
- Would I then be notified if one of the disks failed and went offline?

BigglesPiP · 28 Sep 2009 at 17:25

markyb said:
BigglesPiP, Thanks for the reply.

- Would it be better for me to start again but this time using a software RAID using SuSE?

Yes.

- Would I then be notified if one of the disks failed and went offline?

Lord knows. I know this is an area where a lot of distros are left lacking.

Evil-I · 28 Sep 2009 at 17:31

To be honest software RAID is a little scary with linux, most on board raid is most definately not true hardware raid. I ended up buying an old 3ware RAID card instead which gives me true hardware RAID. My 2 Tb RAID 5 array is made up 6 x 500Gb drives with one running as a hot spare. It's seen by the OS (Fedora 10 in this case) as a single 2Tb volume sdb (I use a seperate small drive for my OS, the array is just data). Theres also a little 3Ware app that monitors the arrays health etc which just basically talks to the RAID cards built in web server. So either CLI interface or browser access.

ThorpedoUK · 1 Oct 2009 at 00:10

You can pickup RAID controllers pretty cheap nowadays, always going to be better than the onboard RAID solution which is, as the guys have said, more or less software RAID anyway.

If you want the job done properly the way it was always intended, there's no substitute for a RAID controller with battery backed write cache.

Any failed drives are handled directly by the controller itself and the OS continues on none the wiser.

For monitoring you can use SNMP, software drivers/tools for your controller and/or software such as Nagios if you need it to be more visual.

ThorpedoUK · 1 Oct 2009 at 00:19

I just noticed your using RAID 0, you'll definately know when one of those disks goes offline or becomes degraded because you will lose the partition outright.

If you want redundancy dont use RAID 0, if you want speed and maximise space with no room for failure, use RAID 0.

markyb · 1 Oct 2009 at 07:03

ThorpedoUK said:
I just noticed your using RAID 0, you'll definately know when one of those disks goes offline or becomes degraded because you will lose the partition outright.

If you want redundancy dont use RAID 0, if you want speed and maximise space with no room for failure, use RAID 0.

Sorry my bad, I meant RAID1. Both drives are mirrored.