HDD Failing

freedomit

Well-Known Member
Reaction score
200
Customer has a HP Microserver Gen8 with 2 x RAID1 and one of the drives is reporting errors. The drive in question is a WD VelociRaptor 500GB 10k drive. The HP diagnostics report the drive as healthly but if i look into the counters one of the drives has a 'Read Errors Hard' count of 102. All the other counters for the drive and the counters for all other drives is 0.

As the drive isnt HP it means im going to have to go though the RMA process with WD which is going to require pulling the drive, testing it, returning it etc so i just want to be sure its an issue before i start the process?
 
They can’t see the drive as it’s on a HP Smart Array controller and in Raid. I would need to pull the drive to do that.
 
Now is a good time to make sure the backup is current and healthy.

I suggest buying a new replacement drive, do the switch, test and RMA the drive in question and set the RMA replacement beside the server as a spare for the next failure.

Yep the entire server is backed up using Windows Server backup to HDD’s rotated daily offsite. Data and SQL is also backed up using our cloud backup for additional protection.

Good idea, I just wanted the avoid buying a spare drive but looks like I might have to for safety.
 
I just wanted the avoid buying a spare drive

Well, your customer is ultimately buying it, and having a cold spare just sitting there for when the next drive fails is a way better scenario than now. Any time we sell a server or a NAS, I always include a spare hard drive in the bid for just such an occasion. Same with power supplies if the server has redundant supplies. Helps me sleep better at night. If that cold spare ever gets used, we immediately order a replacement.

One of my customers has a Dell rack server that boots ESXi from two enterprise SD cards in a RAID1. We have a cold spare of one of those, too. That one actually got used about 2 years in.
 
Well, your customer is ultimately buying it, and having a cold spare just sitting there for when the next drive fails is a way better scenario than now. Any time we sell a server or a NAS, I always include a spare hard drive in the bid for just such an occasion. Same with power supplies if the server has redundant supplies. Helps me sleep better at night. If that cold spare ever gets used, we immediately order a replacement.

One of my customers has a Dell rack server that boots ESXi from two enterprise SD cards in a RAID1. We have a cold spare of one of those, too. That one actually got used about 2 years in.

Yeah I’m starting to realise the importance of having spare. The drive is now EOL and the only ones I can find from disti/reseller are certified refurbished.
 
You don't need an exact replacement (although that is preferred). It's SATA, right? I found this on on Amazon. You just need to make the replacement has at least as much capacity. If you end up getting a non-matching replacement, it's safer to go with a larger capacity so you don't have to worry about the new vendor's 500GB being 50 megabytes less capacity than the original drive.
 
They can’t see the drive as it’s on a HP Smart Array controller and in Raid. I would need to pull the drive to do that.
There is a variant of GSmart that will run on Linux. If you booted from a Linux disk, you may be able to see your raid drives showing up as separate items. I say 'may' because I have had mixed results doing this - sometimes works, sometimes doesn't. Probably with hardware RAID, chances are not good. But may be worth a shot rather than you have to pull the whole thing to bits.
 
You don't need an exact replacement (although that is preferred). It's SATA, right? I found this on on Amazon. You just need to make the replacement has at least as much capacity. If you end up getting a non-matching replacement, it's safer to go with a larger capacity so you don't have to worry about the new vendor's 500GB being 50 megabytes less capacity than the original drive.

It’s a 500GB 10k SATA so not a standard desktop drive and ideally I wanted to try and get one a similar spec. Thanks for the link but doesn’t ship to the UK :(
 
There is a variant of GSmart that will run on Linux. If you booted from a Linux disk, you may be able to see your raid drives showing up as separate items. I say 'may' because I have had mixed results doing this - sometimes works, sometimes doesn't. Probably with hardware RAID, chances are not good. But may be worth a shot rather than you have to pull the whole thing to bits.

This only happens with "fake" RAID cards, basically those that are software based. Being a Microserver it may not have a real RAID card. But, as you mentioned, just booting from a *nix distro will tell.
 
This only happens with "fake" RAID cards, basically those that are software based. Being a Microserver it may not have a real RAID card. But, as you mentioned, just booting from a *nix distro will tell.

One can use passthrough calls (basically a way to talk to the device behind the controller by packing ATA protocol inside SCSI protocol), if available. As I checked, most RAID controllers (either hardware or fake) will do that. On Windows, Hard Disk Sentinel is the software I found working most of the time, and it does see through various controllers (incl. LSI and Adaptec).
 
Last edited:
One can use passthrough calls (basically a way to talk to the device behind the controller by packing ATA protocol inside SCSI protocol), if available. As I checked, most RAID controllers (either hardware or fake) will do that. On Windows, Hard Disk Sentinel is the software I found working most of the time, and it does see through various controllers (incl. LSI and Adaptec).

My point was more to @Mick's observation about booting into Linux, with some RAID cards, not showing the configured array but the individual drives. I learned about this years ago when I first setup my FreeNAS server. Most, if not all, of the low end RAID cards/chipsets are software based and *nix's have no drivers to load the config. So they basically just act as another SATA interface card to *nix's which then use mdadm to build the array.
 
How important is the 10k RPM factor? If you have one failing 10k drive that's both EOL and a type that nobody makes anymore, it may be time to consider migrating that entire array onto new drives.

Newegg does list them as being available new from multiple sellers shipping from the US, so you could always try that route. If you do that and actually do get a new drive I'd consider ordering a second to keep on hand.
 
How important is the 10k RPM factor? If you have one failing 10k drive that's both EOL and a type that nobody makes anymore, it may be time to consider migrating that entire array onto new drives.

Newegg does list them as being available new from multiple sellers shipping from the US, so you could always try that route. If you do that and actually do get a new drive I'd consider ordering a second to keep on hand.

Not huge, the server supports only 3 users but I got them to offer slightly better performance than 7200k drives.

I can get refurb drives from EET Europarts so I might do that as the RMA from WD will probably be a refurb anyway.
 
Sounds like a really old drive at any rate... why take the chance? Swap it out regardless of if you can get an RMA or not.

WD in the end said they can’t RMA it as they don’t make them anymore so have offered a full refund. I’m going to use the money to buy two new drives and rebuild the RAID
 
Back
Top