New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Software RAID and hard drive failure
I did software RAID twice on complete different machines on debian. Both machine end up with hard drive failure. Yes. It is desktop grade hard drive. It works pretty well before and both failed less than 2 months after software RAID was set up.
So, I have to wonder, if software RAID killed the hard drive, especially if it is desktop grade hard drive.
Comments
Not in my experience, I have a rig here as a test bed with WD blues in raid 0 for 3 years
Same experience as Anthony here. I've been running software RAID on some random Hitatchi Death star drives on my backup server and it's been fine.
No.
How would software RAID kill your drives? Desktop drives aren't meant for 24x7 access, that is probably what is killing your drives. The comparison of failures I see between Blue and RE Western Digitals is astronomical in server environments.
Software RAID, especially mirroring or striping, wouldn't add any extra load to the drives. Mirroring just writes data to both drives, striping spreads data out across different drives. So striping actually reduces write by half when you have two drives, but also provides no redundancy.
Some other things you can check:
What types of fans are you running and are they causing excess vibration in your chassis?
What is the max temperature of the drive from the SMART output?
Is your power supply outputting correct voltages?
Did you run extended tests on the drives before you put them in the array?
Linux software RAID has been well known to be very reliable.
I am doing RAID 1 and I am not monitoring SMART. So I don't have a log to say why it happens. Just think it is too much a coincidence. I never had a failed hard drive when it was not running as a RAID.
Blue is good or RE is good?
WD RE4 1 TB Enterprise Hard Drive on Amazon was priced very similar to desktop grade
edit: never mind, I know why it is cheap. It was SATA II.
Even if SMART can't really predict when a drive fails, it still is nice to monitor the metrics of your drives - bad sectors, etc.
I've been running WD SE 3TB drives in my main server off a MegaRAID 9260-8i. They were a little bit cheaper than RE4 drives when I bought them.
even with "Desktop Drive" label marked on disks, Seagate 3T Barracuda is still cheap and solid enough for NAS.
Yes. It is Barracuda. Not sure why. Could be damaged when I have it shipped inside the server to DC. Should pack those hard drive separately.
REs will be better for RAID. SATA II/III doesn't matter, a mechanical drive can't even saturate SATA II bandwidth.
Start monitoring it then. Sometimes it can predict a HDD failure or degradation.
"Monitoring" or not you can always look at what smartctl says for a drive. The log is on the drive itself, so even if the drive is having issues you can ask it what its internal log says.
MAybe the temp killed them
Unless the drive is completely dead and not just having read/write errors.
REs or Blacks are about the same in reliability for software RAIDs. WD RE is the way to go for hardware RAIDs because of TLER. Blues are the low end 7200rpm desktop drives, just as greens are the low end 5400rpm desktop drives. I have seen a lot of Seagate Barracuda drives fail in servers. If you want a reliable drive, go for a WD Black. It will be cheaper than the enterprise drives by Seagate/WD but just as reliable.
As microlinux already said. SATA2 or SATA3 doesn't matter for a HDD, only an SSD. You won't max out the SATAII transport with a HDD. SATAII can handle 375MB/s theoretically, but more realistically about 275MB/s. Most hard drives won't even hit 150.
BIOS is not detecting it
I bought Seagate Enterprise Value. Not sure if I'll regret it.
I've had more issues with WD Black drives than any other. Whether in server or desktop use. So don't get those. But otherwise nothing on the software end should kill a hard drive.
In my experience, Seagate drives failed more than Western Digital drives.
I don't know what the deal is, the last few batches we've ordered (Blacks and REs) have had a 50+% failure rate within a year, they used to be great, but we've stopped using them. Always the same thing, the buzz/beep of death after a reboot.
Strange, all the WD blacks I've had have been very reliable over the years. They are practically the same as the raid editions with a different firmware.
Same, and WD has a much better RMA process. I just compared them the other day, sent two drives to both companies. WD had a new drive to me in 3 business days. Seagate took 8 business days.
We never had issue with Seagate SV35 drives - 4 fails in ~3 years out of ~200 drives.
WD failed more often (non RE), Seagate Enterprise failed more often (surprisingly), WD RE3 failed more often, WD RE4 around the same, Hitachi less.
What's Seagate's warranty? Since one drive is 3 month old and another one little less a year
So you recommend getting Hitachi?
No, as the SV35 are much better priced which compensates this failure rate.
I haven't seen a failed SV35 yet, though we haven't deployed that many of them.
The good thing is they have a 3 years warranty. More than enough.
Maybe I just got a bad batch but I've had 4 out of 24 SV35s fail, 3 of them after the 3 year mark with them only having a 3 year warranty. WD Black and RE have a 5 year warranty.
The main reason I got WD was for the 5 year warranty. Granted I'll get a referb drive if one fails, but that's still kicks the crap out of having to fork over for a new drive.
I am a huge supporter of Western Digital drives. We have tens of thousands of them, and reliability with the RE4's has been superb. Seagate's we avoid like the plague.