Advise on motherboard purchase

[Christopher Chan]

On Tuesday, October 05, 2010 01:11 PM, Jordon Bedwell wrote:
> On Tue, 2010-10-05 at 12:32 +0800, Christopher Chan wrote:
>>> The HDs can operate at up to 70C temperatures but they never get
>>> anywhere close to this as there is enough air circulating inside the
>>> case - at least in my computer(s). As well, the newer HDs run VERY cool:
>>> I have WD 500GB and Seagates and they are barely warm after working all
>>> day. But, of course, they do feel a bit warmer when the ambient
>>> temperature goes up.
>>
>> 70 Centigrade? What have you been taking Basil? I know of disks that
>> DIED at 55/60 (big argument about why the file servers were unstable - I
>> won after I proved it was heat issues heh) so there is no way you can
>> run at 70C which is over 10% OUT of the manufacturers' stated operating
>> environment.
>>
>
> It depends, there are two max temps for an HD, much like there are for a
> processor, you have internal and airflow (external), they usually don't
> differ by much (and shouldn't) if they differ more than 5 degrees you
> have an airflow problem in your case.

Basil is clearly talking operating temperature and not non-operating 
temperature.


>
> Anyways, modern HD's with S.M.A.R.T (all HD's lately, especially large
> external HD's that are not SSD) default to warning at 60c and shutting

I have seen hard disks with a max temp of 55c.


> down at 65c.  They also have what's called the MaxTemp change per hour,
> which a lot of people don't know about,  this is usually 15-20c before
> the HD trips.  MaxTemp change per hour does not include powering down
> for the night, but that does have an effect on the drive in the end
> result.  Spinning down a drive a lot can cause this too and damage a
> drive, which was the topic of major debate on Ubuntu regarding laptops
> if I remember right.

Yeah, whatever, that's why I keep my home box on 24/7.


>
> I would never let a drive go above 50c safely, this isn't because I'm
> worried, it's because I don't want micro-damage, which is especially
> something we worry about on file servers, especially large clusters.  We
> see enough micro-damage, without accelerating.

Nor would I for my mail servers when I was an mta admin and neither 
would the team that was responsible for the file servers but they were 
running FreeBSD 4.x then and there was no smartd on FreeBSD 4.x so they 
assumed that everything was fine nevermind it was a new batch of 
fileservers in a new case model. Only after I got to stick my nose in 
and claim Linux is way faster than FreeBSD at serving files (had proof - 
replacing FreeBSD on my mail servers with Linux got an almost two-fold 
increase in performance at the cost of very slight instability) did I 
prove the disks were running hot. Temps ranged from 40 to 67. Feel free 
to guess which disks died and which did not.


>
> Anyways, if we want to be purely technical, depending on the drive you
> haven't damaged your platter by hitting the MaxTemp stated, you've
> damaged other pieces.  The platters usually take a lot more heat, so you
> could still do a recovery if you have the tools, but that's a whole
> different animal. I think I've covered curie points for platters before
> on this list when somebody brought up HD destruction and recovery.
>
>

I'm not in the business of recovering data so I only care that the 
entire thing is still operational. That means not running anywhere near 50C.