Hi,

 

I setup unRAID about a month ago, precleared all drives, added them to the array, defined the shares and became a happy user. After about 30 days of uptime I noticed one drive became disabled. It had "sde" label. When it became disabled, the label changed to "sdn" and the number of errors increased to couple hundreds. I did the recommended procedure - stopped the raid, turned off the server, changed the sata cable, deselected the drive, started the array without it, stopped the array, selected the drive, started the array and waited for rebuild process to finish. Label became the old "sde" and everything worked fine for 3 days, then I noticed 2 errors and drive became disabled again. After I started the rebuild process it becomes disabled after 20 minutes, and I have to do restart rebuild couple times in order for it to finish.

 

Should I just replace the drive (however the smart report is normal)? Can I just take the drive out and rebuild the array, recalculate the parity (I have about 5tbs of free space on the array and the drive is 1.5tb)? Why label changes and how important is the order of letters? What would happen if I connect this drive to motherboard or other free sata port(now it's connected to HBA card)?

syslog_5_5_13.zip

You can remove the drive and recalc parity based on the contents of the other drives, BUT... any content that was on that drive will be gone. If you want to save the content of that drive you will need to manually copy it to another drive in the array or elsewhere BEFORE you invalidate the current parity. Physically removing the drive and starting the array in degraded mode will allow you to access the parity generated content of the missing drive, but only if all your other drives are good.

 

Drive sd? labels are changeable and aren't used as identification for array drives. The port a drive is connected to probably will change the sd? label, but it won't matter as long as unraid can still correctly ID the drive by serial number.

 

In order to offer specific advice on your drive, you will need to post smart reports, preferably reports on all the drives, not just the one you suspect as being bad. It looks like one of your other drives, disk3, may be in trouble too, but without smart reports it's hard to say.

From what you've described, it sounds like the drive had failed, but you removed/reinserted it into the array and let it rebuild ... which worked briefly.  But it's now failed again.

 

You should do one of two things:

 

(a)  Simply replace the drive with a new drive -- it will rebuild the contents of the failed drive.  This is the best choice.

 

or

 

(b)  Copy the entire contents of the drive to another computer (perhaps an external drive attached to your PC).    Then you can re-initialize the array as you suggested you were thinking of doing ["... Can I just take the drive out and rebuild the array, recalculate the parity ..."].    Note that you could do this copy even WITHOUT the bad drive physically connected 

... that's the whole idea of UnRAID -- the parity disk plus all of the other drives collectively have enough information to recreate the failed drive.

 

I'm not an expert, but it looks, to me, that something odd is happening to both of your 1.5TB drives (sde -> sdn, and sdf -> sdm).  I would check data and power connections to those drives (perhaps they're both powered through the same splitter?).  Are both of these drives connected to the same hba?

 

Also, check your psu - what make/model are you using?

 

In addition, looking at the syslog, I'm suspicious that perhaps one of your software addons is messing up your system (particularly SimpleFeatures).  Can you try running your system for a while with addons disabled?

 

Oh, the order of the drive letters is relatively unimportant - they can, potentially, change every time you reboot.  However, normally, they will not change while the system is running - this would only happen if the drive goes absent, and then returns.

Peter brings up a good point -- with two drives changing assignments, it could be that they're experiencing momentary drop-outs ... either from a "flaky" controller or power issues.  If they are in fact connected to the same controller, you may want to change that;  or if they're using the same power splitter, reconfigure the power.

 

IF you make those changes, I'd remove the failed drive;  zero it on another system (using something like Western Digital's Data Lifeguard);  then re-insert it in UnRAID and let the system rebuild the drive (just as if it was a new one).    That may be all that's needed ... effectively the same thing as option (a) I noted earlier, but with the same physical drive 

Thank you guys for your answers!

 

I would check data and power connections to those drives (perhaps they're both powered through the same splitter?).  Are both of these drives connected to the same hba?

Also, check your psu - what make/model are you using?

Yes, those two drives were connected through the same splitter to different HBAs though... As soon as I opened the case and started to juggle the power connectors I noticed that the splitter was bad and I switched it. However, I started having problems with PSU - it takes like 10 times to press the power button to turn the thing on so I figured it's gonna die pretty soon. I think I will replace the PSU (OCZ 700Watt) and then will go from there.

 

Not really a question for this topic: should I look for PSU with the most number of SATA power connectors or splitters are still not that bad? Which one would do the job for my configuration in my signature (Make/Wattage)?

 

Thanks again everyone!

Get a high quality PSU with a single rail for the 12v line.  Seasonic, PC Power and Cooling, and Corsair are my recommendations.

 

A Seasonic X650 modular supply would be more than adequate - and I understand that Seasonic are very good at providing additional cables on request.  Avoid splitters, if at all possible.