I just installed a new system yesterday, and just this morning one of the SCSI drives failed in my RAID-10 array

5/10/2007 1:31:18 PM

whats the make and model of the drive?

5/10/2007 2:18:00 PM

Dead disks in a raid configuration aren't always the fault of the disks. While it is the most likely culprit when you have a single disk failure there is always the possibility of a bad cable or backplane (if hotswap.)

The thing with real raid controllers is that the controller will take a drive offline at the slightest indication of failure which can include path problems. Once it is "offline" it will not come back online when an intermittent path problem happens to be working as it would when attached to a normal SCSI controller. To bring it online requires rebuilding from the parity data/mirror which normally has to be manually started.

But yeah, you probably do have a failed disk.

5/10/2007 2:42:21 PM

its a Fujitsu Ultra SCSI 320 15k 147GB drive, MAU3147NC

the drive was dead, it was easily identified by the click of death sounds

a half hour rebuild to a new drive and I'm back in business

5/10/2007 2:56:22 PM

got warranty?

5/10/2007 6:23:01 PM

no, but i've got about 342652345645 other drives to take its place

we've got tons of older enterprise class SCSI drives now that we're doing more and more development and testing with SAS/SATA

I just wanted to point out that RAID can actually be useful, had I used a RAID-0 array for performance then my system would have been hosed on the 2nd day. This has happened a lot recently, I think I have the bad touch with electronics

5/11/2007 9:46:24 AM

More than likely it is just the result of abuse on the drives from moving/storing them before you received them. I can't tell you how many times I've had to tell people that yanking a drive out of their RAID array and setting it down on a table while it's still spinning at 10k+ RPM is not good for the drive. ...but I wanted to test my RAID

5/11/2007 12:48:00 PM

Quote :

"no, but i've got about 342652345645 other drives to take its place"

lemme hold some of those drives

5/11/2007 1:22:30 PM

I'll take 3

5/13/2007 12:20:44 PM

raid id a wonderful thing - i wish all or our servers had it

5/13/2007 4:49:03 PM

^^^^ I work at Adaptec, and I get the hand-me-down equipment from our test team. They beat those drives senseless doing stuff like running IOMeter for weeks on end, pulling out drives during rebuilds, etc. Enterprise class drives typically hold up well, but then again if they didnt then we wouldnt be in business I'm not surprised that I've had a drive failure already.

5/13/2007 5:37:11 PM

If you've got 300 trillion drives, lemme get a trillion or so of those.

5/14/2007 1:27:24 AM

^^ I saw a lot of failure analysis reports from a few vendors a few years ago. These were all fibre channel disks from customers who saw high failure rates/repeat failures. Most of them boiled down to mishandling. Thats why drive manufacturers advertise such high MTBF rates, yet consumers never really seem to see the same level of success. They get the crap beat out of them during shipping and then customers start yanking drives out while their still spinning. I think moving them while they are spinning is probably the biggest killer of all.

What type of work do you do at Adaptec? Your stock has been tanking recently. Hope you guys step up your serial storage game this year and bring it back up.

5/15/2007 2:02:07 AM