One HDD 5-10% slower than others on badblocks scan

Setting up a new system using 4 identical refurbished HGST 12Tb drives. As suggested, I am going through the process of burn testing both the system and the HDDs before unleashing any data on the system.

I did the various SMART tests no problem then moved on to the badblocks tests. I have 4 drives (sda, sdb, sdc, sdd) and am running badblocks on each in parallel. I started them sequentially within a minute or two of one another.

Unlike another post on this topic, I made sure that I didn’t have anything running in the background (all prior tests had completed).

Most of the drives are within a few percent of one another (after 65 hours they are roughly done with the second of four write/read tests - 95-98% done with the 0x55 read). For the third drive (sdc) it lags behind by between 5 and 10% (It is currently at 86% of the 0x55 read). It is very gradually falling behind the others (though at a slower rate than initially).

There are no reported errors and nothing obviously different with the drives (as mentioned they are identical models though likely from different batches based on serials - though curiously sdc and sdd are close in serial number)

Any thoughts on why this drive is slower? Once this finishes, I can power everything down swap two of the drives and see if this changes the performance (advice on how best to benchmark appreciated).

From the reporting, the reads and writes on drive C are about 10% slower than the other 3 drives - with a max of 240MiB/s vs 270 on the others.when read/writing the outer tracks.

Is this normal - should I have any concerns about either the controller or the drive?

Could be anything without hardware specs?

Normal hardware variability, I think.

1 Like

Amusingly I just looked up the drives (HGST Ultrastar DC 520 12Tb SATA model HUH721212ALE601) drives and the spec on the WD page says sustained transfer rate of 243MiB, so if anything the other three are exceeding expectations (I assume that 243MiB is if you continuously write to the outermost sectors on the drive.

I’m fine chalking it up to hardware variability (maybe one has more He4 than He3 in it :slight_smile: )

1 Like