Perc 6i: cannot add Hot Spare or do raid level migration

As RAID 0 is not redundant, it is impossible to "rebuild" a RAID 0, so setting the drive as a Hot Spare will not work.  You must "reconfigure" your RAID 0 to a RAID 1, by adding the second disk.

Thanks for the reply - I'm indeed a newbie with RAID controllers, so I'm probably going about it the wrong way.

When you say ""reconfigure" your RAID 0 to a RAID 1, by adding the second disk", how exactly could I do this - is there a link to some info/manuals that could guide me through these steps?

What I'm trying to achieve is to make the C-drive redundant (convert from RAID 0 to RAID 1) by adding a 2nd disk, but I'd like to achieve this without having to delete the VD and starting from scratch.

Most of these functions for PowerEdge servers come from OMSA. I'm not sure OMSA will install on a workstation (it has a different raid management/monitoring solution). I use it for the onboard raid sas controller on my T7500, but I haven't tried it with a PERC6i to see if it offers raid level migrations and things like that.

Edit: as you are new to raid, you may want to check out the article from wikipedia about raid: http://en.wikipedia.org/wiki/RAID.

Dev Mgr has a good point ... this being on a workstation - although the same card as used in Dell servers - has some limitations:

- OMSA (the management application that allows direct interaction with the controller and drives) will not install on a non-server, nor will it install on a client OS (XP, 7), and OMSA Live will not run on a non-server.

- RAID Storage Management is the management application for workstations/client OS's, but, if I remember from another thread, it cannot do advanced functions, such as a reconfigure.

I also do not think this is possible from within the CTRL-R utility.

If RAID Storage Management does not work and it can't be done from CTRL-R, you will need to clone/backup, delete the RAID 0, create a RAID 1, then restore.

Thanks everyone for the feedback.

Indeed, I could not find any possibility to do a raid level migration from within the Ctrl-R utility (hence the post in this forum).

I will read up on (and experiment with) the RAID Storage Management, as I would like to avoid the hazzle of deleting the Raid 0 and starting from scratch again.

It seems that it would be such an obvious piece of functionality to offer, so I'm not clear on why it's not included in the Ctrl-R utility ... maybe an RFC for the good PowerEdge people out there :-)

Thanks again for the info.

I noticed the following in the Storage Manager User Guide (p.25): "You cannot change a RAID 0, RAID 1, or

RAID 5 configuration if two or more virtual disks are defined on a single disk group. (The Logical view tab shows

which disk groups and disk drives are used by each virtual disk.)"

I assume this is case (I have 2 VD's, which are both RAID-0 on the same PERC 6/i controller). However, since I have a 4 TB NAS, it would be easier for me to pump the data over to the NAS, delete the second VD and then attempt to reconstruct the first VD by migrating from RAID-0 ro RAID-1.

What is unclear (could not yet find the answer in the documentation) is whether I will be able to re-create the second VD in RAID-0 configuration if the first VD is now in RAID-1, but I suppose it makes sense that the PERC 6/ would support different RAID levels between separate VD's.

I will continue reading, and then give it a try with the RSM (and report back my findings in the thread).  

You don't have multiple VD's across the same disk group.  You have multiple VD's across multiple disk groups.  Disk group 1 is disk 0, which is also your single-disk RAID 0; disk group 2 is disks 1,2, which is your two-disk RAID 0.  

An example of what it is talking about would be if you had two 1TB disks, configured a 100GB RAID 1 VD across those two disks, then using the additional space, created a second RAID 1 VD across those same disks; you would have two VD's on the same disk group.  This used to be really common, then was discouraged, then brought back again with the availability of large drives (think 6x1TB disks - a single RAID 5 VD would be too large for Windows to handle as a boot drive - it could only use 2TB of it)).

The PERC will support 64 VD's, so you can have as many as you want.  What you do to VD0 will NOT affect VD1.  When you put different VD's on the same set of disks is when you run into problems.

Are you also going to add a drive to VD1 - is it really a RAID 0 too?  If this holds your user data, you should know that there is no redundancy to a RAID 0 ... if one drive dies, it's all gone.

Storage Manager User Guide (p.25): "You cannot change a RAID 0, RAID 1, or RAID 5 configuration if two or more virtual disks are defined on a single disk group. (The Logical view tab shows which disk groups and disk drives are used by each virtual disk.)"

Yes, you can create multiple arrays at different RAID levels on the PERC 6/i. That section of the guide is referring to sliced arrays. A sliced array is when you have more than one array across the same disks. For example, if you have 2x1TB HDD's and you had a RAID 1 that used 500GB of space and a RAID 0 that used the remaining 1TB of space across those drives.  That would be an example of a sliced array. It is not possible to reconfigure a sliced array. You do not have sliced arrays, so you don't have to worry about that.

Here is the section regarding the reconfigure process from the OMSA 6.5 user's guide. It is listed as Virtual Disk Task: Reconfigure:

Thanks

OK - got it, thanks for the explanation on the disk groups.

Did not yet have time to experiment with the RSM, but will plan this in for the weekend.

Indeed, VD1 is a 2-disk RAID-0, it's main purpose being to serve as disk space for video editing. As I'm using (relatively) cheap SATA 7200 rpm drives (instead of SCSI disks), I'm using RAID-0 to maximize the disk throughput and to reduce the risk of I/O limitations when running back and forth on the video timeline. 

I'm aware that there is no redundancy, but since I'm running daily backups against the NAS, I would never lose more than a days worth of work in case of a disk crash.

Hello Daniel,

from theflash1932, I understood that OMSA cannot be installed on a workstation, especially not when running a client OS such as Win7. So, this would leave me with the choice to try to install RSM and try the raid level migration with this software. Correct?

I'm aware that there is no redundancy, but since I'm running daily backups against the NAS, I would never lose more than a days worth of work in case of a disk crash.

Ok ... sounds like you've given it some thought.  Some people just think RAID 0 = better performance, then when they lose a drive, they are disappointed that the RAID system wasn't better :)

Hello Gammelman

OMSA uses IPMI to communicate through the BMC to the hardware of a server. Since workstations do not have BMC's it would not likely work. The SAS RAID Storage Manager was designed for our SC servers and workstations. It is the Openmanage for non-BMC systems.

So, yes you would want to use RSM. OMSA web server can be installed on client systems to allow remote management of servers, but local administration via OMSA on a workstation would not likely work. I notice that I posted a link to the reconfigure instructions for OMSA in my previous post and that is probably where the confusion is coming from. I should have posted the instructions from the RSM, sorry about that.

Thanks

Thank you for setting me on the righ track - I downloaded RSM as per your link, and used the Reconstruction Wizard to add the second physical disk and change the RAID level to RAID-1 from RAID-0 (in RSM, I selected the VD and then from the menu; Operations->Advanced Operations->Reconstruction Wizard).

The computer is operating normally and VD0 is now a 2-disk RAID-1 volume.

But, there were some hickups;

** 1.) the log files tells me that there were quite a few of Fatal Errors, examples below:

[FATAL, 3] TIME = 19-01-2012 07:09:48 LOCALIZED MESSAGE = Controller ID:  0   Puncturing bad block:   PD   --:--:1      Location   0x3973c191

[FATAL, 3] TIME = 19-01-2012 00:13:10 LOCALIZED MESSAGE = Controller ID:  0   Reconstruct detected uncorrectable multiple medium     errors:       ( VD   0  Location   0x71dd99e,       PD   --:--:0Location   0x71dd99e)

** 2.) At approximately 52% of completion, there was an unexpected power outage that shutdown the computer and so interrupted the re-configuration and raid level migration process half-way. However, the computer rebooted normally, and the recontruction process started automatically in the background.

TIME = Not available LOCALIZED MESSAGE = Controller ID:  0   Reconstruction resumed:   VD       0

TIME = Not available LOCALIZED MESSAGE = Controller ID:  0   Reconstruction started:   VD       0

**3.) But, in the log files I find new Fatal Errors:

[FATAL, 3] TIME = 20-01-2012 14:05:27 LOCALIZED MESSAGE = Controller ID:  0   Background Initialization detected uncorrectable multiple     medium errors:       ( PD   --:--:0  Location   0x13ec58bb  VD       0)

[FATAL, 3] TIME = 20-01-2012 14:33:49 LOCALIZED MESSAGE = Controller ID:  0   Background Initialization detected uncorrectable multiple     medium errors:       ( PD   --:--:0  Location   0x221db9c7  VD       0)

My follow-up question is - should I worry about any of this? Any action I should take?

Hello again Gammelman

There is nothing you can really do about these errors. There were bad logical block addresses on the drive. When it reconstructed the drives into an array it tried to recover these blocks, but since the previous configuration did not have any redundancy there was nowhere to get the correct LBA data from.

These are corrupt blocks of data across the array that are not repairable. I would recommend running a consistency check across the array to make sure all of the bad blocks are marked off. If these LBA's contain any important file information then you will encounter issues with the OS or program the file is associated with.

After you run a consistency check I would recommend a chkdsk /r to repair any file system corruption. If you encounter any issues with software on the system then i would recommend uninstalling and reinstalling it.

Thanks