I got my 10x HDDs delivered from ServerPartDeals, they’re Dell branded Exos drives, Seller Refurbished
I know one of the first things I’m supposed to do is make sure they’re configured to run at 4K, not 512, so I’ll be checking that.
I’ll also plug them all into a HDD dock and read the SMART data off them.
Afterwards, I remember reading a guide (here I believe) about how to test all the drives. There was a command line tool to check the drive for errors.
Does anyone know what I’m talking about? I remember I had to run a plugin on the computer to make sure I could run multiple instances of it.
Alternately, I intend to spin up a TrueNAS instance for this, does TrueNAS have any built in tool to fully check the drives before I create any pools?
when I get new hdds, i open a tmux instance with one pane per hdd, then i do:
- smartctl -t short /dev/sdX
- smartctl -t conveyance /dev/sdX
- badblocks -b 4096 -ws /dev/sdX
- smartctl -t long /dev/sdX
And if it passes all those, then I put them into a zpool.
edit: for bigger drives, you might need to increase the block size for badblocks, 65535 for example.
Can you tell me why you go for 2^12 but then choose (2^16)-1 as an alternative? Wouldn’t that create alignment issues?
I would use CrystalDiskInfo on them.
I usually just yolo and throw them into the pool but my last drive I just did a long SMART test and it came up clean so into the pool it went.
One can also do a badblocks run if you don’t mind waiting the extra time to do so and can understand the logs.
I was going to post this same thing.
Currently have it running on 2 - 20 TB Exos X20 with the defaults.
Looks handy!
Would that default 8096 block size be large enough for 18TB HDDs, or would I have to go higher?
I think 8192 is high enough, since 4096 only breaks after something like 16TB.
Why would it “break”? Isn’t the 4096 the block size setting?
a) If you plan to encrypt the drive, VeraCrypt “full format” when encrypting will fill the whole HDD with random data. If there’s a bad sector or corruption, VeraCrypt will let you know or you can use CrystalClearlnfo application simultaneously.
b) If you’re not planing to encrypt it, do the “complete format” option on Computer.
c) Any application that overwrite with one and zero (just search for them) and then use CrystalClearlnfo.
Each one of these options can take a long time to finish, a 5TB HDD can take 12 hour in each of these options.
Crystaldiskinfo, disk drill, numerous other programs.
I use this very handy shell script that automates multiple badblocks runs.
i run a surface test with Hard Disk Sentinel. it takes a couple days but it’s a good peace of mind assurance.
I just toss them in
you’re going to run truenas? so ZFS is gonna be your file system?
If so, here’s what I do (I use ZFS)… plug 'em in, run a SMART check (just in case DOA), then BAM straight into production.
No bullshitting.
I have had HDDs pass badblocks, & still fail ZFS. Seeing as it doesn’t get much more strenuous than badblocks, why tf should I bother if even bb isn’t weeding out the weak drives? I have back-ups, so I have nothing to worry about. Just straight in, ZFS will sort 'em out.
If you’re using ZFS, dont’ bother with all the fluffing about of testing etc. ZFS *IS* the test.