Hard disks spinning down then spinning up again

[Joe Fenton]

After an update a few days ago on my Freenas 9.10, I can now hear the hard disks almost constantly individually spinning up all the time. This never used to happen, and I'm sure won't be doing the disks much good. Can anyone give any pointers to what I can look at to stop it, or is it a bug?
Ideally I would like to use powersaving as the server is often doing absolutely nothing through the day.

 

[DrKK]

If you do the math, spinning down your hard drives, in 2016, rarely provides any substantial power savings, and worse, the $0.72 per month you save is offset by $6 of wear-and-tear to the hard drive itself.

I strongly recommend just going in to the disk settings in the GUI, and turning off all spin-down/power-saving features of the drives. All of my drives are "always on", with all power saving features deactivated, and the ENTIRE freenas is only consuming about 35W, including the drives.

It's a fool's errand, really, trying to "spin down" the drives. You are literally better off, with modern, NAS-centric drives, just letting them go. That's what they're designed to do.

 

[Joe Fenton]

Thanks very much for that, I assumed the cost would be a lot more. Although energy prices are probably higher in the UK, it's still not much.

 

[depasseg]

Where is your System dataset location configured?

 

[Joe Fenton]

Just had a look as I didn't have a clue, and it just says 'files' which is what the disk volume is called. The only other option in the dropdown is the boot device (a flash usb drive)

 

[depasseg]

So you are writing system data to your "files" pool. That happens periodically and is likely the reason the drives wake up. You can move the system dataset to the boot pool and see if there is an improvement in sleep/wake. But as DrKK mentioned, if you only have a handful of drives, you aren't going to save much power.

 

[Joe Fenton]

Is that bad? Should I be storing that data on the boot pool anyway? Will a failure mean something is not recoverable?

 

[depasseg]

Is that bad?

Not necessarily

Should I be storing that data on the boot pool anyway?

It's up to you. In this case it might make sense if you'd like to limit your main pool access.

Will a failure mean something is not recoverable?

Yes, your reporting graphs and log data will be lost if your boot pool is lost (it's suggested to mirror the boot device to reduce this risk/impact).

 

[DrKK]

Yes, your reporting graphs and log data will be lost if your boot pool is lost

That's a pretty god damn good point, that I never thought of before.

But of course, my syslogs are on the pool so it doesn't affect me.