Had my parity drive fail tonight, with a red ball and a large amount of errors. Switched off my server and checked all the connections and created a smartctl report. Please find it attached, along with a syslog.

 

I am running unRAID 6.0-beta3

smart.txt

syslog.txt

Wholly crap Batman! Look at the number of reads on the parity disk. If it were bytes it would be 18 Exebytes, but I think each read is a 4K sector, making if 4096x more. Must be a mistake. I think it might take lifetimes to read that much data!

 

For what its worth that smart report looks fine. No reallocated sectors. But with Seagates I'm always nervious, as I've had some do flakey things even with no hint of a smart problem.

 

My first thought it maybe a memory error? Would explain the read errors and the garbled read count.

Heh, yea I was pretty sure that read count was an error.

 

Well the drive is currently in the process of recreating the parity info with no errors so far. When I have some time available where I can switch off my server I'll run memtest (I had run it for 24 hours when I first built the server) and see if anything pops up. Hopefully there's no silent corruption going on

The smart.txt looks ok from my view, however, you've never done a smartctl long test.

 

I would suggest you stop the array and take it off line for a little while and run a long test on the drive.

It will take a long time. It will give you a "line in the sand" for measurement.

I.E. an entry in your SMART log on the drive.

 

The other choice is to turn off the spindown timer for a while and run the test.

 

You can't have any activity to the drive during a long test or it will abort.

 

I always run the smartctl -t long on new drives so I get a base point in time for the long test and the hour it was run.

Then once in a while I do it again on each of my drives to insure there aren't LBA's that are getting weak.

Thanks for the advice, I'll run the long smartctl test tomorrow and see if anything looks off.