I’ve read through the Disk Management documentation here: Disks | TrueNAS Documentation Hub . It left me with a couple of questions:
Scheduled SMART Tests: Is there any reason not to run a scheduled full SMART test on the drives in a pool?
1.1. I know performance will be degraded while it’s happening, and since I have 8 disks I might actually try to stagger it so only 4 at a time are tested (one from each of my 4 vdevs) on different days; but
1.2. Am I correct that it’s not somehow dangerous to do automated SMART testing in TrueNAS?
1.3. I’ve had failing disks fail SMART tests before, so I want to be running scheduled LONG tests on my production NAS unless there’s some reason I shouldn’t be that’s specific to TrueNAS.
Power Management. Putting aside for the moment the question of whether spinning the disks up and down is a good idea at all …
2.1. The defaults are ALWAYS ON and no Advanced Power Management. I admittedly am completely unfamiliar with the Advanced Power Management modes beyond what’s written in the docs.
2.2. Are there any guides to when and how to adjust these settings for, e.g., a home office NAS? I’m not committed to doing this, but using less power is always good, especially on this system, which tends to idle around 85 watts with the drives inserted. On the other hand, this also isn’t the sort of thing I’d want to blindly experiment with.
Thanks!
EDIT: This discussion is relevant to how to actually spin drives down and schedule SMART checks using cron. Leaving this here for later reference: HDD Sleep/Spindown/Standby
Thanks for this reminder. I’m new enough to ZFS that this would have slipped my mind entirely.
I need to figure out how long the scrubs take.
A full SMART test on one of my drives takes about ~12-14 hours, so I’m probably going to set it to run overnight.
I’ve got 8 of them, in a pool with 4 mirrors, so I’m thinking about scheduling two weekly test runs with 4 drives each, so there’s always a drive in each mirror not being tested.
That way there’d always be one disk in each mirror that wasn’t being tested, in case a lot of activity hits the pool. That seems like it might be the better way to go as far as maintaining a better minimum level of performance during the test.
Is that worth it, or am I overcomplicating things?
On the client side, I’ve got hourly TimeMachine backups scheduled. I’m sure my disks will love trying to do SMART tests while those run. I need to figure out if I can adjust the TimeMachine schedule…
Depends mainly on how much data’s in your pool and drive performance. My scrubs take about 2 days for ~80 TB of data on spinners (4, 6-wide RAIDZ2 vdevs).
I schedule all my SMART tests at the same time–short tests daily and long tests weekly. I then make sure scrubs start at least 24 hours after the long SMART test is scheduled.
Scrubs are scheduled for weekly: Sunday at midnight.
I scheduled the long SMART test on all my spinning disks for 10PM local time on Wednesday. It takes about 27 hours, so it should be finished well before the scrub starts.
I only have 50 TB available on this pool (8 disks, 4x mirror vdevs), so if your much larger array took 2 days, I should be fine.