Do S.M.A.R.T. requests wear off the harddrive?

In a german ubuntu guide for the hddtemp program is claimed that every S.M.A.R.T. request wakes up the drive although no data is read. Because some harddrives that can park their head into a parking position can do so only a certain number of times and will wear off earlier. I never heard or read about this before.

I wonder if the claim is true at all and why the head has to be put out of its parking position when only SMART is requested. The statement that only certain drives can actually park their head confuses me more, how do I confirm that my drives do or not? And will the once per minute request that hddtemp does really have a noticable effect on my drives? I mean they will die sooner or later anyway.

I'd also appreciate more sources on this topic than just a statement written to some wiki.

Baarn

Posted 2012-01-13T16:23:24.460

Reputation: 6 096

2I don't imagine parking the head has any effect on how long the drive lasts. Spin up cycles on the other hand can kill a drive if there are too many, but is rare this happens anymore. – Moab – 2012-01-13T16:39:44.887

1Just to increase the confusion, why would they read smart if the drive has been put on standby ? My highpoint controller has smart capability, yet the drives go to sleep quite nicely still, but I do not regularly monitor its temps via the raid manager. and now that you mention it, the one drive I have in smartfan being temperature checked , I dont think it goes to sleep. Very interesting Question. – Psycogeek – 2012-01-13T16:48:59.410

1> Spin up cycles on the other hand can kill a drive if there are too many... I have read this often, but never understood how exactly does spinning up the spindle could wear the drive out any more than any other activity. > but is rare this happens anymore. I had not heard this part. So—counterintuitively—modern drives are more sturdy and can withstand being turned on and off several times a day/several thousand times in it's lifetime? If so, then there is no reason not to turn a system off when it's not going to be used for a while. – Synetech – 2012-06-07T22:21:31.863

hmm, Speedfan seems to get SMART data really quick, it doesn't look like it makes many requests at all. Almost seems like the data's just there for it! Does hddtemp do SMART differently?! – barlop – 2013-01-18T16:19:13.683

@barlop the problem is, that although no data is read directly from the platters, they do spin up. I guess the SMART data is available before they are completely spun up, as it is stored on the controller of the drive. – Baarn – 2013-01-18T16:29:31.110

Answers

The problem you describe is real. It is true that most hard disks will break down once a certain number of head-parkings are done.

For an official news as regarding at least the Western Digital's line of Caviar Green hard disk drives, see this article : WD Caviar Green HDDs Suffer from a Critical Design Flaw, which discusses the not-so-intelligent wonderful new IntelliPark feature of WD and how it destroys disks.

A long thread discusses this problem : WD Green Drives - BUYER BEWARE, started by someone whose new WD Green Drive failed too soon and who discovered this problem, as well as a solution (for WD only).

As to the question of when does any hard disk park its heads, you will have to search the supplied documentation and/or ask their Support. The operating system can put a drive to sleep and wake it up, but doesn't normally control the head-parking. This is done by the disk firmware, and may or may not be parametrable by a utility of the manufacturer (as is the case for WD).

I would assume, however, that a sleeping disk will park its head, so too many sleep cycles are to be avoided. A sleeping drive can probably not answer SMART requests, so insisting upon one will wake the drive up.

But again, different drives may behave differently.

harrymc

Posted 2012-01-13T16:23:24.460

Reputation: 306 093

Thanks for your answer, I am awarding you the bounty because I think it solved my problem and helped me dig up this thread (although I am not on a laptop): https://forums.gentoo.org/viewtopic-t-694832.html

– Baarn – 2012-01-29T23:33:54.127

All hard drives park their heads at some point. It's part of normal hard drive operations. Normally hard drives park their heads when they experience excessive G-forces, or are being shutdown.

A hard drive needs to move its head out of the parked position in order to read and write to the platters (to read / write data).

That some hard drives can only park their heads a limited numbers of times seems erroneous to me.

user978122

Posted 2012-01-13T16:23:24.460

Reputation: 429

Hard drives do have a limited number of contact start/stop or load/unload cycles. For example, consumer desktop class Seagate Barracuda hard drives are rated to 50,000 or 300,000 cycles depending on capacity; see the data sheet for details.

– bwDraco – 2012-01-13T18:11:30.557