Stress testing new hard drives

[owlig11]

Is it best practice to stress test new hard drives? I have (2) 14TB WD Red Pro drives that I plan using in a NAS. Ive read online that some people run "badblocks" in Linux, while others use the "Preclear" add-in in Unraid.

What is the general consensus here, do you stress test new hard drives before use?

 

[Steve]

I would not say so. Once upon a time when drives were far less "autonomous" than they are today that made sense, since those older drives were unable to look after themselves. But today's drives are able to deal with common issues transparently. So there's really not much reason to worry about any new drive's state.

 

[PHolder]

Experience with RAID has told me no. Granted these are all the same brand and size, but the results over 30 drives or so over the years has never had a drive fail early in its life. I presume this is because the drives are tested during manufacturing before shipping to a customer, and because the technology is such that modern drives are built around failing and recovering at the best of times. At the densities we've achieved, there is no such thing as a perfect drive surface any more... so the drive firmware uses lots of error recovery technology just to exist.

 

[SeanBZA]

In general, with new drives, I do the SMART full surface scan. Yes takes a good number of hours to run, but at least does verify the factory formatting actually did work. then occasionally will run it again, to verify again all sectors are readable.

 

[rfrazier]

It's been a few years since I've had to buy a hard drive. Recall that drive WRITES from an OS on a normal PC are NOT verified. My practice in the past was to fill a new drive with random gibberish then run a SR level 4 on it. Then delete the gibberish. Yes, I would do that even on an SSD. With modern drives in the multi TB capacity range, there's no chance the manufacturer has done a surface analysis on a HDD, or storage cell analysis with an SSD. If nothing else, this "brings the controller's attention to each sector" as SR says in its own documentation. If there are physical or mechanical problems or major surface defects, they would show up. Obviously, some of that doesn't apply to SSD's. But, even then, it gives each storage cell a test. I'm not saying this is a right or wrong answer, maybe just shades of gray. But, I'd probably do the same thing now if I bought a new drive. I Run SR level 2 yearly even on SSD's. I also try to leave 10 % empty space so the drive can do its housekeeping. For myself, I also turn off write caching on the HDD in case of a power failure or computer lockup.

Ask 50 engineers a question and you'll get 200 answers.

May your bits be stable and your interfaces be fast. Ron