interesting unraid failure

[porina]

Of course, when it comes to data storage, interesting is a bad thing. I was doing a routine backup when I noticed a little red X on one of the drives. I poked around the web interface and apparently there were write errors. I did a short SMART test and that reported back ok, and the SMART attributes were ok too.

 

I'm about to see if I can salvage a spare disk from an old server as a temporary replacement and let the array rebuild. The 4 drives in the backup server are all WD reds, and my old server used non-NAS Toshiba drives. While not ideal, it's still better than nothing.

 

Once extracted I'll connected the suspect drive and run a full test on it. The machine that houses the unraid is a HP Microserver Gen8 so I would hope it is not failing in some way.

[porina]

WD's own tool doesn't report anything in SMART, but using another tool I see the raw value for seek error rate is non-zero, which I don't think is normal. Doing a long scan on it now. I also note the other hard disk in the system I connected it to (old WD green) is showing pending sectors so that's probably on the way out too...

 

With the replacement drive in the server, it is estimating 21 hours for a rebuild... maybe I should seriously look at double-redundancy.

[samiscool51]

bad SATA cable or port?

never seen a failure like that....

it's always a good idea to have double-redundancy

[Blake]

On 11/6/2016 at 4:51 AM, porina said:

WD's own tool doesn't report anything in SMART, but using another tool I see the raw value for seek error rate is non-zero, which I don't think is normal. Doing a long scan on it now. I also note the other hard disk in the system I connected it to (old WD green) is showing pending sectors so that's probably on the way out too...

 

With the replacement drive in the server, it is estimating 21 hours for a rebuild... maybe I should seriously look at double-redundancy.

I have had something similar a few times, but it was also affecting the read speeds, was the sata cable.

 

If you have them in a RAID, most raid cards don't pass through the SMART status. (I know the gen7 onboard doesn't)

[porina]

7 hours ago, samiscool51 said:

bad SATA cable or port?

never seen a failure like that....

it's always a good idea to have double-redundancy

 

1 hour ago, Blake said:

I have had something similar a few times, but it was also affecting the read speeds, was the sata cable.

 

If you have them in a RAID, most raid cards don't pass through the SMART status. (I know the gen7 onboard doesn't)

It is a HP microserver so everything is set up already. I could reseat the master connector on the mobo but there's not much I can do about the rest.

 

It took about 24h but the rebuild with another drive went smoothly and I have a working array again. The suspect drive passed a full surface read, then a full surface write so far. The SMART value for seek error rate has gone back to zero again now. I'm hesitant to put this drive back in the array, given how long it takes to do a rebuild. So I'll reuse it elsewhere as an ancient WD green is now showing SMART warnings with pending sectors.

 

I would add more redundancy but the enclosure has 4 bays and they're all used. I looked for systems with more drive bays but haven't identified one yet at the right price/performance I'm looking for.