Reallocated Sector Counts.

[Zithras]

I bought 8 new 1.5 TB Seagate drives (with newest firmware) to start out my unRAID servers.

 

Of these, 1 was obviously bad, showing thousands of errors, 98 reallocated sectors, and slow access times.

 

I have just formatted the discs and am currently running a parity check (about 300 minutes remaining).

 

5 Discs seem perfectly fine.

 

1 Disc had a single reallocated sector after the formatting.

 

The final disc had 7 reallocated sectors.

 

Are these last two discs okay?  I know having bad sectors on brand new discs is a bit weird, but should I RMA them, or just keep an eye on them and make sure the count doesn't increase too quickly?

 

Thanks,

Zithras

 

(Can post more disc data if requested to)

[SSD]

The number of realloxated sectors is not as much of an issue as the whether the number holds steady or not. I recommend running a couple parity checks a week for the first month or so of a new array. If the reallocated sector counts hold steady you are fine. But if, after each check the number goes up, that would point to a problem disk.

[Zithras]

Unfortunately, the reallocated sectors on those two drives are now 62 and 10 (after the parity check).

 

Still, I'll keep them as long as they don't go up, since the computer doesn't seem to have trouble accessing data on them like on the bad drive.

 

Thanks for confirming that I don't need to be too worried (yet).

[SSD]

I don't want to be too reassuring.  Most disks have 0 bad sectors.  1 or 2 is not uncommon.  But anything higher is unusual, and if the drive starts to get more and more errors that is a very bad sign.

 

Run another parity check, if the counts go up again, RMA them.

[Zithras]

Ah.  Will do - thanks

[Joe L.]

Unfortunately, the reallocated sectors on those two drives are now 62 and 10 (after the parity check).

 

Still, I'll keep them as long as they don't go up, since the computer doesn't seem to have trouble accessing data on them like on the bad drive.

 

Thanks for confirming that I don't need to be too worried (yet).

Do you have any sectors that are "Pending Reallocation"?  Those would be sectors that were unreadable, but since nothing has since written to them, they will not be re-located until a write occurs.

 

I would not worry about those counts on a 1.5TB drive unless they continue to climb with use.  Odds are as good as any they will stabilize once the initial set of flaky sectors are handled during the first few parity calc/check processes.  Fortunately, the odds of a bad sector having any of your data is far less when the drives are initially empty.

 

Did you perform any burn-in on your drives before you added them to the array?  The preclear_disk.sh scriptI wrote  is designed to exercise a disk and expose the poor performing sectors (as well as anything else) before you start using the disk for your data.  I would strongly suggest you use it when you grow your array. (It is probably too late now to use it, since it sounds like you have added the drives to your array configuration)

 

Joe L.

[Zithras]

Ugh.  One of the other two is bad (at least, the reallocated sector count seems to be increasing) - and I might as well RMA the other just in case.

 

That 3 of 8 seagate drives bad out of the box.  Is this normal?

 

I'll be sure and use the preclear script on the new drives.

 

Doesn't formatting and parity checking expose these bad sectors anyway?

 

Update: The second bad drive had 76 (and growing when I stopped the second parity check) reallocated sectors, and 7 pending sectors.

 

Zithras

[Joe L.]

Doesn't formatting and parity checking expose these bad sectors anyway?

Sectors are marked as "Pending Reallocation" when they cannot be read.  Sectors are only re-allocated when they are subsequently written to. 

 

Formatting writes very few sectors. That is why it can occur in a minute or so.  Writing to the entire drive takes many hours. (About the same as a full parity calculation)  Parity checking reads all of the sectors of every drive.

 

The preclear_disk.sh script performs a pre-read of all the sectors on the disk in a linear fashion as well as random reads and continual reads of the first and last sector.  In other words, it bangs the heck out of the drive moving the disk head far more than usual. 

 

After the pre-read, the entire disk is written to, and it is then followed by a post-read, similar to the pre-read, in that it again tries to keep the disk heads moving back and forth randomly across the disk reading random sectors as well as reading every sector linearly from first to last.

 

The script does a smart report before it begins, and another at the end, and shows you what changed.  It has been very interesting to read of the results from various users.  Most do report some number of reallocated sectors with the 1TB and 1.5TB drives.  Occasionally, the reallocated sector count keeps growing as the drive is used.  For those, an RMA is in order.

 

Good luck with your RMA process.  Seagate has had a lot of issues with firmware lately... Hopefully, the new drives will have the updated firmware.

 

Joe L.

[Zithras]

Thanks - while I'm waiting for the new drives, I'll go ahead and run the preclear script on the existing drives - it will at least give me a baseline to monitor from as time passes.

[Joe L.]

Thanks - while I'm waiting for the new drives, I'll go ahead and run the preclear script on the existing drives - it will at least give me a baseline to monitor from as time passes.

You will need to un-assign them from the array in order to run the script on them.  It will not let you clear a disk that is assigned in your array.

 

Joe L.