Safely separate a RAID0 without destroying data?

[Salty Wagyu]

Intel RST has been throwing up errors today stating that Volume0 has failed and Disk on port 1: failed. I've looked at most of my folders on the volume and everything appears intact and working, chkdsk reports everything is in working order as well. This is a raid0 array of 2x 1TB Samsungs.

So what I want to do is check the SMART status of the problematic drive with DiskInfo or such, which you can't do whilst it's in an array. What's the recommended method to go about doing this? Would breaking the array to read the SMART data destroy the contents of the disk?

[n_K]

You can't check SMART status? I haven't ever seen or used intel RST but in device manager (Windows?) does it show the actual drives under 'storage' or 'hard drives' at all? If so you might be able to check the SMART status or is there an intel tool for configuring the RAID controller you can use?

[carmatic]

there is a utility called smartctl in the linux live bootable cd which sounds like exactly what you need... assuming that your RAID controller can pass the SMART data ... otherwise you'll have to unRAID the hard disks before booting the live linux cd and running the utility, and then reenable the RAID if you want to use your computer normally again

[Salty Wagyu]

Have not been able to no, from what I've read on google this is typically normal behaviour for a raid array. DiskInfo and HD Tune never was able to view any SMART data of my array. The volume is shown in Device Manager

Running a verification test at the moment, going to be a while

there is a utility called smartctl in the linux live bootable cd which sounds like exactly what you need... assuming that your RAID controller can pass the SMART data ... otherwise you'll have to unRAID the hard disks before running the utility, and then reenable the RAID if you want to use your computer normally again

And unraiding & re-enabling the raid0 won't affect the data?

[n_K]

It's bad advice to unraid and reraid a volume, different controllers do things differently e.g. PERCs store the RAID details in the controller, HPs store it on the drive but cache it in the controller and verify it on bootup and other cheaper RAID controllers might just store it on the drive, so if you unraid it and it deletes data near the beginning of the drive... Not gonna be able to get it back.

[Xilo]

Didn't know you could disable raid, boot to livecd, then reenable raid without issue.

RAID0 is never a good idea...

[carmatic]

And unraiding & re-enabling the raid0 won't affect the data?

it really depends on your controller, like other people have said ... if worst comes to worst you might just have to boot into the live cd and use another utility which creates an image of your volume, and save that image on a good hard disk in case the re-RAIDing destroys your data.... in which case you will boot back into the live cd and use that utility to restore your volume

otherwise, just back up your data onto a good external hard disk in case the erroring hard disk really does break down...

[Chris123NT]

I actually just went through something similar, you can unraid and re-raid in the RST BIOS utility, and the data will remain intact. BUT you need a program to restore the partition table. I used testdisk which you can get here: http://www.cgsecurity.org/wiki/TestDisk

Both of my arrays recently lost the meta data for the array (Controller update boned both RAID0's). Unraiding, re-raiding and using this program saved my bacon and all of my data was perfectly intact.

[Karl L.]

My understanding is that since RAID 0 involves striping instead of mirroring, if one disk fails then you irrevocably lose all of your data. Since you seem to still have access to all of your data, your first step should be to back up everything as soon as possible. That said, its possible that Intel RST is trying to indicate problems SMART flagged, such as too many bad sectors or insanely high read or seek error rates, instead of all-out failure. Since I have never used Intel RST I cannot guarantee that this will work, but I have never had any problems reading the SMART status of individual disks using GNOME Disk Utility or smartctl, even those in a soft-raid (mdadm) configuration.

[Salty Wagyu]

Thanks for the advice. Will need to think about my options now. My raid0 is backed up daily to a 2tb drive in another machine, it's just the downtime, waiting for a drive replacement, then copy stuff back that's a headache.

Already had a failed samsung earlier this year, so I bought a new one retail and turns out it was causing a lot of vibration to my case, claimed a refund on that when the RMA'd drive came back and rebuilt the raid. Can't believe my luck with another drive fail this year.

I enjoy the fast loading times in games from having raid0, which I can't use an SSD for as my games library is about 300GB+. Just need to decide on whether to go with a single WD Caviar Black 2TB drive (and pray it's quiet) or get another 1TB and rebuild raid.

[carmatic]

with your luck with hard disks right now, i would think that raid1 would be a good idea, the read speed should be pretty much the same as raid0 on an intel rst controller which is what matters for loading games... problem is you need two 2TB hard disks, it will be more expensive but it means failed drives would only slow your computer down while you wait for a new drive to arrive , you have to do almost nothing on your part to keep using your computer ... hopefully the single one thats working in the meantime stays working until you have the new drive to rebuild the array

[Salty Wagyu]

Tried another app today - HWiNFO32 which is able to read the SMART data, but I'm unsure if it's accurate and don't know what to make of the values either, anything unusual? Intel RST hasn't reported any errors for 2 days so far.

[Japlabot]

12 Posts and nobody has said this? Take a BACKUP dude. It doesn't matter what you do but have a BACKUP first. It's only a 1.8TB Array. 2TB Drives are cheap. Worst case scenario you break the RAID and lose all the data, no problemo, just recreate the RAID and restore.

[Salty Wagyu]

12 Posts and nobody has said this? Take a BACKUP dude. It doesn't matter what you do but have a BACKUP first. It's only a 1.8TB Array. 2TB Drives are cheap. Worst case scenario you break the RAID and lose all the data, no problemo, just recreate the RAID and restore.

Thanks for the advice. Will need to think about my options now. My raid0 is backed up daily to a 2tb drive in another machine

:)

[Open Minded]

just recreate the RAID and restore.

If both drives were good (new), then yes. Since one is giving problems, I'd trash the broken drive, buy an SDD (for the OS) and use the one good drive as a storage/program drive. RAID 0 was a great option for maximum speed before the days of SSDs, but now just buy one drive and be done with it. Or push the limits and stripe a few solid states...

I think you're missing out and setting yourself up for future problems if you rebuild the array. Use the last good drive for storage. I have a 640 WD Black drive set as D:\ so when I install programs that can't fit on my little 60 gig SSD, I just change the install path to D. There's also a shortcut to my music folder which is stored on the D drive. Works great and I've yet to have a problem.

[Salty Wagyu]

New drive arrived today, and decided to run them separately now. DiskInfo's SMART is showing a caution for the drive Intel RST warned about in "Current Pending Sector Count" 252/252/0 (Current/Worse/Threshold) Ran a chkdsk with /R and it found 4kb in bad sectors.

It failed Seatools Short DST test and recommended running Seatools for DOS to fix the bad sectors, it also passed SMART test, hopefully I can skip this bit and is reason enough to get a replacement from an RMA.

p.s. It didn't look like I could safely separate the raid0 anyway, the Intel Matrix Storage Manager in BIOS menu warned deleting a raid or resetting disks to non-raid would delete data, but didn't matter at this point. Annoyingly switching from RAID to AHCI proved pointless too, my SSD with Windows on it would bsod.