Hello lovely hivemind (:
i try to build a more power efficient setup.
my HDD (ST16000NM001G) draws 7watts or even more if my power meter is correct.
according to the specs it is 7200 rpm. And now i wonder if it is possible to set it to 5000 rpm or so to save power. It would obviously become slower but that is ok.
I know that HDDs don’t like to be spun down and back up again which causes wear.
Is the same true for permanently lowering the rpm?
And if not how to do it?
You cannot, nor should, lower the RPM of a drive.
Spinning down the drives is fine unless you spin them up dozens of times every day.
IMO there are a few difficulties here…
Because you don’t want them to keep spinning down and up, you need to profile your disk activity and decide what times of day you can spin them down and them NOT to spin up again for several hours. You may need to adjust the timing of all your automated tasks to concentrate them on the hours when the disk is spinning. (So if you have solar photovoltaics, you probably want to have them spinning during the day when you are generating electricity and in the evenings when you are watching media, and then have them spun down whilst you are asleep. If you don’t have solar, but you have cheap night-time mains electricity, then the opposite might be true.
You then need to find a way for spin down to happen only for the periods you want BUT for the spindown to happen reasonably quickly once that period starts.
You probably still want SMART monitoring for the disks when they are spun up - and (for the moment at least) it seems to be all or nothing if you stick with supported SCALE functionality.
There is a 3rd party truenas-spindown-timer script which I looked at which spins down whilst still enabling SMART, but when I looked at it I didn’t think it met the above needs, and it needed splitting into two scripts - see Provide a cut-down alternative script designed to run from Cron for an extended discussion on what I think is required.
I don’t think that it would be that difficult to chop the existing script into two parts:
You also need to make one change to the SMART service definition so that it doesn’t check spun down disks and wake them up unnecessarily.
A couple of caveats:
ixSystems have (IMO quite reasonably) said that running a script like this puts you outside support on e.g. SMART issues. They also have said IMO quite unreasonably that having a script on your disk BUT NOT ACTUALLY RUNNING also puts you outside support (go figure!!).
I have never managed to get my I/O even in the dead of night down to zero. Unless you can achieve this, there is little point in trying to spin down the disks. And this is why I: a) abandoned the idea of using these scripts, and b) didn’t bother hacking together the two scripts I had the idea for.
thank you for your insights.
i guess the thought to save power this way is gone by now 
are the advanced power management settings worth sth?
No idea. My disks are set to 128.
If you can get disk usage for extended periods down to literally zero (with SMART service OFF), then you can probably make this work and get some savings. So I wouldn’t give up quite yet.
is 128 better than the default disabled ?
I have no idea. My suggestion is to try it and see how the power draw changes.
It is, likely by a marginal amount, but it is.
The marginal power savings in dollars in cents will likely be offset with reduced life of your drives and having to purchase new ones sooner. Drive spindown/head parking has a significant impact on endurance.
I believe SMART lists this as Load_Cycle_Count, and it is fairly well known that higher values of this correspond with higher failure rates.
Side story: I used to run a bunch of iMac labs. After a string of failures, and having to take them apart with suction cups and a heat gun…I would intentionally prevent them from ever spinning down in software.
that’s why i wonder Level 128 says minimum power but without spin down
it reads like it would save power without added wear on the drive
That is a data center drive and as such explicitly meant to not be spun down.
What you should do instead is change/disable the default period for head parking, which is set to just a few minutes. The latter makes sense for a data center but for home use it means unnecessary wear and tear. I don’t know the command for Scale, but Google or someone else here can certainly help.
Lastly, your wording implies that you have only one single HDD. If that is the case, please change it asap.
The user manual is here. Section “2.5.4 Extended Power Conditions - PowerChoice™” seems to me to be encouraging users to save power (and hence money) by spinning down, and states that (only) spindown timeouts of < 15mins “may limit this products reliability and data integrity”. Section 2.1 states that the drive is rated for 600,000 spindowns (over the 5 year warranty period) so that is 120,000 spindowns per year or 328 spindowns per day - and since there will be only 24x4 = 96 spindowns per day with a 15 minute timeout, this does NOT seem to be an issue.
i bought it based on this backblaze report
yes in this setup it is the case (:
since its there to replace my previous one which had 6x4TB WD Red with RAID-Z2.
The Old setup will become a “cold storage” which only boots once a week to pull replication from the “live storage” which is the mentioned disk.
And for important data there are also cloud syncs in place (:
this is indeed a very interesting read o.o
i am currently also investigating the proxmox server to maybe get the cpu to save more power than just setting the cpu governor to power save.
but the HDD pulls nearly 50 of the whole rest of the setup ^^
i am willing to lower some performance of the setup to save power but i would rather not wear down the drive unnecessarily (:
Level 128 is your best option if you don’t want to spindown the drives.
If you want to spindown, consider the following script.
I have already referred the OP to the ngandrass truenas-spindown-timer and to my extensive discussion with the developer about what scripts are needed to a) get the spindown idle period calculated correctly and b) vary the schedule to meet your specific circumstances.
This is not to say that the existing scripts don’t work - IMO they are fine if you want the same constant spindown timeout at all times. But if you want to vary them by ToD or DoW etc. then some variant scripts simply derived from ngandrass’ excellent work are needed as mentioned above.
Sorry, still adjusting to the new UI.
So that is 24x 7 = 168W per day ; 1.2 Kw-hour per week, so 63 KwH per year.
At 0.10$ / KwH, that means about 6.30$ per year.
Say you manage to half your power consumption, so you “save” about 3 dollars per year.
Should your disk costs 90$, you will need to save for 30 years before paying back a single drive.
Considering how this kind of spin up and down wears disks down, you will not save any money by spinning your drives down. You will actually increase your cost as the cost of increased wearing is higher than the cost of power…
sadly i am closer to 0,41$ / kWh 
And i am on the cheapest renewable energy i can get.
only one cheaper is nuclear explosion of the planet wich would save me around 2ct / kWh …
Ok ; so 4 times more expensive, so 4 times shorter for the same saving, so 7.5 years instead of 30. Still, a hard drive rarely last that long.
The concept remains the same : the saving from power will not pay back for the extra wearing on the drives.
To break a lance in @DS_DV favour, modern NAS HDDs are designed to handle a few dozens spinups per day of their warranty period: a couple each day won’t likely impact the drives lifespaan.
