Re: [gentoo-user] New hard drive. Is this normal? It looks like a connect problem.

[Michael <confabulate@kintzios.com>]

[-- Attachment #1: Type: text/plain, Size: 6312 bytes --]

On Saturday, 10 May 2025 16:53:55 British Summer Time Dale wrote:
> Dale wrote:
> > I didn't know about that until now.  I already shutdown my old rig. 
> > Might try that later.  It may shed some light on this mess. 
> > 
> > I did send a email to the seller tho.  They sell a LOT of drives.  I've
> > seen them show a stock of over 200 drives of a particular model and a
> > day or so later, sold out.  They sell new, a few kinds of used as well. 
> > I tend to buy used but most of the time, the number of power on hours is
> > in the single digits.  The recent drives show 2 hours each.  I think if
> > it is a problem, they will know since they test a lot of drives.  Maybe
> > it is normal but if not, I'm sure they will agree to swap or refund. 
> > They sold out of the 20TB drives shortly after I ordered mine.  They
> > started with right at 200 and sold out in like 2 or 3 days. 
> > 
> > I figure I'll hear back shortly.  They been pretty fast to respond to
> > questions in the past. 
> > 
> > Dale
> > 
> > :-)  :-) 
> 
> I got a response.  This is what they said. 
> 
> > Thank you for bringing this to our attention. As long as we're not
> > seeing any I/O errors that would inhibit your ability to use the
> > drive, everything should be fine.
> > 
> > This type of link speed negotiation issue can occur with helium-filled
> > drives, as their spin-up time tends to be slightly longer than that of
> > traditional drives. Is your system or HBA a bit on the older side?
> > Most modern toolsets and software account for this extended spin-up
> > time by allowing a longer delay before attempting speed negotiation,
> > which typically avoids this issue altogether.
> > 
> > In summary, this isn't unprecedented behavior when working with older
> > hardware or software, but at this stage, it doesn’t point to any major
> > functional problem. I hope this information helps.
> 
> As I mentioned, it passed all the SMART tests.

What do you get for the smart attribute with ID 22?

https://www.backblaze.com/blog/smart-22-is-a-gas-gas-gas/

Although others report ID 16 as the "Current Helium Level", or "Internal 
Environmental Status" attribute.  The ID number and Attribute description 
depends on the drive firmware.


> I'm not sure on the
> 3GB/sec connection yet tho.  I'm pretty sure that mobo is capable of
> 6GBs/sec tho.  When I put it in my main rig, I'll know for sure. 

Slow spin-up or not, if it is not performing at 6Gbps as advertised when 
connected to a SATA 3 bus, then it is not fit for purpose - assuming transfer 
speeds are a consideration for you and you don't want to let this slip.


> What are your thoughts on what they say?  It make sense to anyone who
> knows more about hard drives than me?  Now if they can just find that
> last drive I ordered that is several days late. 
> 
> Thanks. 
> 
> Dale
> 
> :-)  :-) 

My knowledge of drives is quite limited and my working knowledge of large 
Helium filled drives is a fat zero.  Despite this, here's some random thoughts 
- should you wish to read further:

I have read drives which have seen continuous service in large datacenters and 
crypto-mining farms for a couple of years are decommissioned, tested, reset to 
zero and sold cheaper as 'refurbished'.  If you keep an eye on Amazon and 
other large retailers and you notice large batches of refurbished drives 
suddenly show up sold at cut prices, then this is in all likelihood their 
origin and explains the low prices.  When you check the perturbations in 
supply you'll notice some makes, models and sizes of drives arrive rather 
prematurely compared to their age in the refurbished drives marketplace and 
this is an indication of early failure rates higher than the big datacenters 
were wishing to see.  It doesn't necessarily make all of these drives bad, but 
it is something to bear in mind when you check how much warranty they are 
being sold with after they are labelled as 'refurbished', compared to the 
original OEM warranty when new.

Regarding Helium sealed drives, they are reported to have a slightly lower 
average failure rate than conventional drives.  Helium having a lower density 
than air and not smelling anywhere as bad as methane ;-) is used to reduce 
aerodynamic drag of the moving parts within the drive.  The idea being such 
drives will consume less energy to run, with less windage the platters vibrate 
less and therefore they can be packed tighter, they will run cooler and at 
least theoretically will last longer.

The laser welding techniques to seal the helium in the drive casing and keep 
denser air out is meant to ensure the 5 year warranty these drives are sold 
with when new.  In practice, any light weight small molecule gas can leak and 
in this case the drive will lose its Helium content - and soon fail smart 
tests.  As it loses Helium at some point it will start to draw more energy to 
operate in a higher drag environment.  Since any SATA controller power 
threshold is not unlimited, the increased drag will cause a slower spin-up 
than when it was new.

I'm not saying your drive is failing, but the slow spin-up argument *because* 
...  Helium, could be somewhat moot.  Modelling studies have shown ceteris 
paribus a Helium filled drive will spin *faster* and remain cooler than an air 
filled drive:

https://www.researchgate.net/publication/225162945_Thermal_analysis_of_helium-filled_enterprise_disk_drive

You can check if smartctl output shows a different Spin-Up Time value against 
other drives - if this Attribute is reported at all.  The Average Latency of 
your 20TB Helium filled drive is reported in its data sheet as 4.16ms - the 
same as 16TB, 14TB, 12TB non-Helium Ironwolf Pro drives.  This indicates the 
time for an I/O request to be completed, not necessarily a spin-up performance 
alone, but why should your 20TB be slower to spin up?   I don't know.  :-/  

Anyway, these are a lay person's comments.  A drive engineer will know exactly 
what's what with this technology and its performance variations.  A chat with 
Seagate's support may get you closer to the truth and explain why the 16TB 
drive spins up nicely while the 20TB drags its feet.

HTH,

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]