HDD or SSD 'problem' threshold... what's yours?

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
So you get a SSD from ebay, you install it to test it before using it...

It only has 3 power on hours, but has 1 re-allocated sector and 10 soft errors, the drive in question is an Intel Enterprise drive. You decide to format it (long) and see if any error # increase.

- If no additional errors do you keep the drive?
- Do you RMA/RETURN the drive no matter what, any errors or relocated sectors = instant return?


What about for a HDD that you plan to use in RaidZ2 or RaidZ3?
- Do you only put in HDD that are perfect, or are some relocated sectors OK?


I'm just curious what everyone's threshold is personally at home, and if you can share at work too I would find that interesting. As I keep adding more and more drives to various systems, and overall qty I'm finding it harder to find 'perfect' drives, or rather 8 perfect for this system, and 8 perfect for that system, etc... Obviously, my new drives if they're not perfect I will return but I'm talking about used and/or refurb. from ebay.


I personally am running some HDD in RaidZ2 that have a couple relocated sectors, and same with SSD. As long as the # doesn't increase after doing a full-disk (long) format and then doing some writing/reading and benchmarking I'll run them.
 
  • Like
Reactions: Son of Homer

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
Hard drives often will have some reallocated sectors but I don't view a few some time in history as an issue and should not increase when you test them prior to use.

I have no real opinion on SSD, I have not generally seen any with errors that I have looked at, I guess I would be a lot less happy to se any errors increase on a SSD, I expect and SSD with errors increasing on your initial data read/write tests probably has a short life ahead. (I don't know if this opinion on SSD is sound or not)
 
  • Like
Reactions: Son of Homer

mervincm

Active Member
Jun 18, 2014
159
39
28
IMO perfect was never even a manufacturers' goal. If they are stable, I use them. If I see them increasing or performance impacts from "marginal" areas I dump them.
I gave up trying to predict if / when a drive would fail, so I don't do any proactive replacements based on age either. I tend to keep a lot of copies of my important data, and even a couple copies of the not so important data.

IMO Stable is important, Perfect is not. :)

SSDs have always been stable till they completely fail in my experience. I don't have much experience with enterprise SSDs,
 
  • Like
Reactions: Son of Homer

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
What made me think of this thread was I recently got a half dozen enterprise intel SSD and all of them have barely any power on hours, but all have identical or nearly identical "issues" such as: relocated sectors (1), soft errors (10), and read/raw error 39. I just found that a little strange.

After doing complete formats of two, as well as 20GB+ of benchmarks and writing no errors increased so I'll be running them :)
 

mervincm

Active Member
Jun 18, 2014
159
39
28
Maybe the oem set them up a little non-standard on the SMART side? It's just a guess but when I see a pattern I don't expect I try to remember to consider if its my expectation that's "wrong"
 

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
Or maybe that's a unique error list due to a firmware update or similar and is completely normal. Since you have more than your one like that And tested ok i am going to guess that it's ok :)
 
  • Like
Reactions: T_Minus

acquacow

Well-Known Member
Feb 15, 2017
784
439
63
42
Flash either works or it doesn't usually, so I just format and go.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
Looks like you're probably right about the Dell smart data, everyone has identical 'issues' and usage handful of reboots and less than 4 power on hours. Probably some weird/funky smart... It's not in a dell, so just a normal HBA in IT mode.

I've gone through 3 of them so far and after a full 400gb format, and then 8-20gb of writes 0 change in any smart data, so that's awesome :)
 

pricklypunter

Well-Known Member
Nov 10, 2015
1,708
515
113
Canada
Spinners I'll only put into use after a battery of extensive testing, any problems show up and I'll replace them, same for SSD's. Re-allocated sectors are indicative of further problems further down the road. In your case though, it's unlikely to be an issue with the disk itself, given that you have a batch of them showing the same oddball smart values. I would be happy to use them, providing no other problems are showing up with test data :)
 
  • Like
Reactions: T_Minus

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
IMO perfect was never even a manufacturers' goal.
Actually, "logically perfect" media is what hard drive manufacturers provide. Any media bad spots have been mapped out and hidden by the manufacturer when the drive is built. Any errors you can see are the result of things happening after manufacture. You can't tell if the reallocated sector is the result of something innocuous like a write splice error or from the drive being dropped and the head hitting the media.

I've never had a manufacturer complain about an RMA for even a single pending bad sector.

I won't put a drive into service if it shows any errors.
 

i386

Well-Known Member
Mar 18, 2016
4,221
1,540
113
34
Germany
I test my ssds & hdds with full format in windows, run vendor specific tools (when available) eg intel ssd toolbox and look at the remainin lifetime. If I can't see this data (smart data wiped/manipulated, or like some dell ssds with proprietary/custom smart data) I won't use it.
 
  • Like
Reactions: T_Minus

mervincm

Active Member
Jun 18, 2014
159
39
28
a lot of this comes down to your tolerance of a failed drive. Home lab, I keep multiple copies and very rarely do I use an HDD outside of a redundant array. The impact to me is at most a slower performing array while it rebuilds, this leads to me being more tolerant of less than perfect I suppose. In the enterprise we replace most everything at the prefailure warning stage, but our 3pars will keep a disk with some bad blocks without calling for a warranty replacement. Even in the enterprise there is a precedent for using known less than perfect :0
 
  • Like
Reactions: T_Minus