I got a Netapp DS4243 with 24x 1TB for Christmas. I currently have it connected to my Dell Poweredge R730 through a PERC H830 in HBA mode. My server is running Proxmox and I’m primarily using it for a media server and a Pterodactyl server.
There’s probably some way to do this from the UI these days, but I tend to prefer the command line. It’ll depend a little on the topology you want, but for a pool of striped mirrors, you’d do something like:
zpool create pool-name mirror /dev/disk1 /dev/disk2 mirror /dev/disk3 /dev/disk4 ...
Here’s the relevant documentation for OpenZFS. Use the -o flag to include any options you want (you probably want to include -o ashift=12 and I usually use -O compression=zstd so that my base dataset includes compression—the capital O indicates it’s a feature for the dataset rather than the pool). If you prefer the UI, check the PVE docs.
So if I understand correctly, this…
zpool create JBOD_DS4243 -o ahsift=12 -O compression=zstd raidz2 sdaa sdab sdac sdad sdae sdaf sdi sdj raidz2 sdk sdl sdm sdn sdo sdp sdq sdr raidz2 sds sdt sdu sdv sdw sdx sdy sdz
should create a pool with 3 raidz2 vdevs with 8 drives each?
That should do it. Different people have differing opinions about how you should identify the disks (I like to format them with a GPT partition and then use the GPT label to note which bay they’re in and the serial number; that way when one fails, it’ll be easy to tell which is which).
Having said that, this way is just fine. Make sure you do catalog which disk is where in a spreadsheet or something. The /dev/Ada style letters can change, so you want to know which serial number is in which bay.
Edit: Oh! And note that you’ve got a typo in ashift before you copy and paste. Looks like I did too in my original reply. Just fixed that.
1 Like