I’m getting a delayed write failure on the D drive at work, which made me lose the data from one of my analysis runs. Ran chkdsk and there are all kinds of bad sectors. Have a new drive on the way from Dell.
Problem is, our IT guy is having open heart surgery so I’m kind of on my own. The D drive is a RAID 0 setup with two SCSI drives (the C drive is the boot drive). How can I tell which one (or both?) has the bad sectors? I understand what RAID 0 does, but that’s about it.
If I unplug one of the SCSI drives, will chkdsk run on the one that’s left or does the “D” drive not exist at that point?
RAID 0 does striping of the disks so you have 2 disks but really they look as one. If you pull one drive in the RAID 0 set it will no longer exist. You are describing RAID 1 or mirroring… Do you have them confused or is it actually a RAID 0?
jeff… raid 0 sucks… it’s actually a misnomer because it provides no redundancy… and is actually useless.
bad sectors don’t mean a drive is shot, but in your thought process it and scenario, it is (because you can’t replace the drive by simply swappign them out)… you are never going to be able to replace a drive in a true RAID 0 format wtihout losing all data in the drive set… that’s why it sucks, it stripes two drives but doesn’t give any overhead that mirroring or raid 5 sets do, to allow single (or multiple) drive failures but retain data.
in windows disk manager, how do your logical volumes look? let us know or take a screenshot
also what raid card do you have? you might be able to put the new drive in and add it to the existing configuration, but the issue is that the array stripe is across the ‘bad sector’ drive as well as the ‘good’ one.
have you tried a good old fashion chkdsk? was it an online or offline run? as in, did you reboot and have the machine inaccessible as the process ran?
If my analyses have to solve out of core, it can do a lot of reading/writing to the drives, so I was hoping the SCSI and RAID 0 would help with that.
Everything is backed up, and I can still access the drive. It was just some of the new files that failed when writing them. CHKDSK was repairing stuff, but it was still running after 16 hours and someone needed to get on the computer. Maybe it would have fixed all the bad sectors, but Dell decided to send a new drive.
Start making backups! As Sonny mentioned, RAID0 is not forgiving when it comes to recovering/restoring data.
If you are getting an identical drive (I hope you are…) then I can think of a possibility where you can just plug and play it. You could make a drive image of the bad drive from another computer (or from some bootable media without the other drives attached). Acronis makes a nice program that WD and Maxtor use in their tool set. It might do the trick? This is just a possible concept to try.
Have a backup of your data anyways!
I had a RAID0 boot drive fail on me not too long ago. It only had a few important files and was backed up occassionally. My credit card payment receipts were the worst thing I lost.
I back up my model files probably 4 times a day, and sometimes have 10 or more progressive versions of the model.
I don’t care about losing everything on the drives, so do you think guessing which one is bad, replacing it with the new one, and running CHKDSK would work?
With the programs you’re running: YES, RAID 0 does have a larger performance advantage with it.
However, if you’re running it with RAID 0 you are on the razor’s edge if one disk dies.
If you need to save time with large out of core simulations, then RAID 0 is still the way to go. If you want to avoid this in the future, then do a RAID 1 or other mirrored setup.
Also, IIRC, with RAID 0 you can’t just pop it in, you have to do some other things like set up the stripe set in the RAID BIOS and what not.
his issue with raid 0+1 or 1 is hardware… if you only have two drives, you’re kind if in a pinch, you mirror or have optimal i/o throughput… the raid 0 splits up the disk writes across hardware… so the speed is there, its’ just that it relys on the stripe and controller and is garbage for redundancy… since jeff has the backups and this is essentially swap space for him, i’d say this raid is optimal (especially since it’s really the only other option than mirroring with two drives)…
yeah, you’d have to put the drive in and write the stripe from controller memory… that’s why i was asking what hardware it was, some lower level deals write the stripe to hidden disk partitions, that when it gets corrupt, is all shit. i’d boot to the raid controller and see how the controller addresses the drives, whether in degraded or failed status… then go from there.
Jeff, 16 hours is a lot but that process is pretty slow… for that disk space i’d say thats about expected… maybe a little less than optimal. We planned on 36 hours for 1tb chkdsk on a pretty hefty controller.
So what we’ve learned here is that I’m a piss-poor backup for our office’s IT needs, and it’ll take me probably 4 times as long to accomplish a task. But at least I have a charge number for it!
All I can see from the system configuration sheet is that there is a Poweredge expandable raid controller in PCI slot 1, and I think it uses an Intel® 631xESB/6321ESB Ultra ATA Storage Controller - 269E.
I’ll give it a shot and see what happens. I need to wipe all the data off the drives anyway since I have to send the old one back to Dell.
