Fixing Onyx’s RAID 1 failure

Topic

New Reply

HARDWARE DIY By Will Fastie Before I was able to decommission Onyx, my daily driver desktop PC, one of my hard drives failed. Hard. Why the hard failu
[See the full post at: Fixing Onyx’s RAID 1 failure]

5 users thanked author for this post.

cloudsandskye, Norio, fisher75, DLivesInTexas, rc primak

Reply | Quote

Replies

While I do not have a RAID configuration, there are only two brands of HDDs I will accept for mission-critical applications, including using them as backup drives: Seagate and Western Digital. And not all WD drives. I have not had an outright HDD failure, but the Toshiba drives just don’t seem to perform as well as WD or Seagate. This is true of internal and external drives.

For SSDs, I have no complaints about the Samsung EVO series, or the Western Digital Blue series. (WD Black would also be very good.) I don’t trust most other brands, even where the internal components are the same part descriptions, manufacturers and numbers. Inland has also not caused me any issues. I have had one off-brand Phison-based SSD fail after just one year.

 

-- rc primak

2 users thanked author for this post.

rbailin, Will Fastie

Reply | Quote

Having two new drives file almost immediately suggests a handing error, which I’m sure in your case isn’t you! You don’t say how you obtained them (Amazon, etc.) but I’d suggest those two drives had been fairly brutally treated somewhere along their way to you.

FWIW, I’m running 5 8TB Toshiba NAS drives in a Drobo. The Drobo is a repeated source of problems (crashing, losing drives, overheating) but the drives themselves have been solid for a couple of years. Seagate on the other hand: I don’t think I’ve ever had a Seagate drive which hasn’t had problems. I’ve got an almost new external one now which has a strange SMART error, although all tests, including Seagate’s own, say it is fine.

(I also have a QNAP, and that is – touch wood – reliable, with WD CMR drives in it.)

1 user thanked author for this post.

Will Fastie

Reply | Quote

rc primak wrote:

but the Toshiba drives just don’t seem to perform as well as WD or Seagate

There were Toshiba and Hitachi drives in my IBM ThinkPads over the years, and they performed well. After my Toshibas failed, I asked around and heard more negative sentiment.

Reply | Quote

Graham wrote:

Having two new drives file almost immediately suggests a handing error, which I’m sure in your case isn’t you!

Thanks for the confidence; I don’t think I abused them. As for handling in transit, it’s certainly possible but I’ve bought a lot of drives via mail order over the years and this is the first time I’ve experienced out-of-the-box mortality.

I did buy the drives through Amazon. I also bought the FireCuda drives that way and was surprised to find two boxes inside the outer Amazon box – both marked NewEgg.

Reply | Quote

could have easily been shipping, handling, even a prob in the factory like mfg temp. Murphy lives! 🙂 Very unlikely you.

Reply | Quote

It’s possible the FireCuda seller was a third party seller who sells on both Amazon and Newegg.

Reply | Quote

Considering the tolerances built into high capacity drives it is amazing that they work as well as they do. Everyone seems to have their preferences for drives. For my Synology NAS I have 4 4TB Seagate Ironwolf drives. I have it set up using Synology SHR instead of RAID 5. This way I can slowly replace the 4TB drives with larger capacity drives and actually be able to use some of the extra space without having to upgrade all the drives before enlarging the array. SHR used in that way is still fault-tolerant. The Ironwolf drives have been solid for three years so far. I also had a RAID 1 array in my previous PC that used the predecessor of Intel RST. During it’s 10 years of mostly continuous use I had two hard drive failures that were easily recovered. For the record, one was a Seagate Barracuda and the other was a WD Caviar. Everyone that has experience using RAID emphasizes that it is not a backup solution. I make sure that I have external drives for data backup and I’m looking at adding a cloud drive service for an offsite backup in case of disaster.

1 user thanked author for this post.

Will Fastie

Reply | Quote

We set up automatic backup routines to Amazon’s S3 from our QNAP drives (work and home) and that’s been flawless. Between RAID and that cloud backup, I’m feeling pretty confident, although I’m thinking of syncing my NAS to an external SSD as well.

Reply | Quote

I was surprised to see Fred’s negative take on RAID. While I’ve never used RAID on my desktops, I definitely do on my QNAP NAS (at home and at work). The work one uses RAID 5 and I opted for RAID 10 on the home one. Both the work and home NAS experienced a failed drive at one point and the hot-swap replacement worked like a charm. I can’t imagine not using RAID on a NAS, although I’m considering switching from RAID 10 to RAID 5 for my next QNAP at home.

Reply | Quote

Lisa Horn wrote:

I was surprised to see Fred’s negative take on RAID.

In Fred’s defense, he started writing about this nearly 15 years ago. Back before software/firmware-based solutions were widely available, a physical RAID controller board was required. That was definitely a complicating factor.

And although Intel’s software has improved, I think it’s still safe to say that there is a bit more complexity involved than with single, SATA-attached drives.

Reply | Quote

I have never understood/bought-in-to the need for RAID. I have had computers from RS (little ones including coco), Commodore, IBM PC, and perhaps 10 of my own custom built boxes and a couple bought ones, and a smattering of laptops, and I have never lost data due to storage failure (stupidity, yes, but hardware failure, no). I back up fairly regularly.

Back in the days, there seemed to be a tad of validity for going that way.. But hardware reliability has vastly improved. Yes, Murphy is alive and well, but, still – a few backups here and there and done. Safe. (terabyte, full drive, image. Easy and quick.)

I always viewed it as an unnecessary added complexity, and complexity in any electronic system, even in the early IBM 7070’s, is a bad thing and WILL beget trouble.

I sit aghast at people who lust for it.

EDIT: yes, brands are all important. Samsung for SSD and mostly WD for spindle with a couple Hitachis mixed in there that never failed. I blew a 650 Ti Nvidia card once – fried it, but not a hard drive. I don’t keep them long enough.

 

Reply | Quote

a wrote:

I have never understood/bought-in-to the need for RAID.

I have a DIY NAS with a RAID 10 array, used primarily as a file server.  With the RAID 10 array, the same movie can be viewed from three or four devices simultaneously without flicker, stutter, or drop in frame rate.

Always create a fresh drive image before making system changes/Windows updates; you may need to start over!

We all have our own reasons for doing the things that we do with our systems; we don't need anyone's approval, and we don't all have to do the same things.

We were all once "Average Users".

Computer Specs

Reply | Quote

a wrote:

I have never lost data due to storage failure

I have, and it hurt. I’ve always done backups, but unless backups are created after every change in files, something will be lost if the drive fails.

Since I started mirroring, I’ve lost nothing.

a wrote:

I always viewed it as an unnecessary added complexity

A point for Team Fred!

a wrote:

I sit aghast at people who lust for it.

Just to be clear, I don’t lust for RAID. But I do long for long retention of all those photos.

Reply | Quote

a wrote:

Very unlikely you.

Other than leaving them in the box for nine months, I don’t think it was me, either.

Reply | Quote

Then we get in to “were the heads parked correctly”… but assuming it wasn’t left in the sun and fairly well temp controlled they should have been fine. Factory fault.

Reply | Quote

Thank you for writing a useful article about RAID 1.

My question:
I assume the RAID drives are connected to a RAID controller in the Onyx computer.  Or are the drives connected using software RAID?

If a dedicated RAID controller was used what model of RAID controller is recommended?

Thank you for your help.

Reply | Quote

It’s using the onboard Intel controller. This is standard on a lot of computers these days.
Use whatever controller is on the motherboard, no need to buy more hardware.

cheers, Paul

Reply | Quote

cellsee6 wrote:

Or are the drives connected using software RAID?

It’s not hardware. It’s a combination of a UEFI setting, Intel firmware as part of UEFI, and an Intel driver in Windows.

Reply | Quote

The new FireCuda drive has been running for 225 hours now, about 10 days.

How long do you think I should let it run to have confidence that it won’t be a victim of infant mortality?

Reply | Quote

One year.

Reply | Quote

b wrote:

One year.

Good research, but I can’t wait that long.

Taking that as a yardstick, the average US life expectancy is 78 years. My rule of thumb is five years for a hard drive. So if we take 1/78 of five years, we come up with .06 years or roughly 560 hours. That’s just over three weeks, which is the time I suggested I’d wait in the article.

I admit to some anxiety. I want to install the second FireCuda drive before the old Baracuda drive gives out.

1 user thanked author for this post.

b

Reply | Quote

Sorry but I can’t find the reference for an article I read over the weekend suggesting that Seagate Exos drives were preferable to Segate Iron Wolf drives on a cost per GB basis plus you get extra features. Most Synology NAS boxes support these drives.

Examples on Amazon:
Exos X16 14Tb 7200 RPM 256 MB Cache – $209.99 (per GB approx. $0.0164)
Iron Wolf Pro 16Tb 7200 RPM 256Mb Cache – $279.99 (per GB approx. $0.0171)

Pretty close but you also get a drive designed to run 24/7 and other features.
I’m certainly giving it consideration for when I replace my Iron Wold Pro’s.

May the Forces of good computing be with you!

RG

PowerShell & VBA Rule!
Computer Specs

Reply | Quote

RetiredGeek wrote:

drive designed to run 24/7

Which is designed for 24/7 operation – Exos or Iron Wolf?

Reply | Quote

Exos Enterprise level drive.

May the Forces of good computing be with you!

RG

PowerShell & VBA Rule!
Computer Specs

Reply | Quote

RetiredGeek wrote:

Exos

I’m gonna need to read the spec sheets.

Reply | Quote

RetiredGeek wrote:

Sorry but I can’t find the reference for an article I read over the weekend

Exos vs. IronWolf Pro – Which is the best HDD option for your NAS? (YouTube).

Reply | Quote

Will,

You nailed it. It wasn’t an article but that youtube video.

May the Forces of good computing be with you!

RG

PowerShell & VBA Rule!
Computer Specs

Reply | Quote

I was wondering if the article had an error in the paragraph following Figure 5. Under Physical Devices, the drive with ID 0 is not part of the RAID array. I think the new drive has ID 1. Regardless, this was a very interesting article. I was never interested in implementing RAID, but this article piqued my interest.

1 user thanked author for this post.

Will Fastie

Reply | Quote

cloudsandskye wrote:

third party seller

An interesting thought. The drives were shipped from Amazon, but the seller was a third party. But that doesn’t really explain the NewEgg boxes. It’s possible that the seller is an aggregator, who buys excess inventory and resells it, meaning that they might have bought stuff that NewEgg had already packaged.

Reply | Quote

cloudsandskye wrote:

the drive with ID 0 is not part of the RAID array

Correct. That’s drive C:.

But what I was trying to say is that the new drive was shown as a “member” of the RAID volume with ID 0.

Reply | Quote

I just replaced a failed drive in a RAID 1 on Windows 11.  Everything happened in the OS.  Nothing happened in the UEFI.  I replaced the failed drive, booted the system, went to the Intel software and told it to rebuild the RAID, and that’s it.

It’s hard to picture how this could be made any simpler.  You can’t get around having to physically replace the drive.  I guess the Intel software could automatically rebuild the RAID or proactively ask whether it should rebuild the RAID.

Reply | Quote