Rebuilding Truenas questions about Pool and vdevs

I have dual identical Nas boxes running TrueNas Scale.

Each one has 128G of ECC ram.

16 Iron wolf Pro 12tb drives (room for four more in this chassis)

AIC card 4 1tb NVME drives.

Dual 256G NVME Mirror Boot

Dual Intel X710 SFP+ 10G

My Data is my life, I cannot replace it ever?

Monthly I back it up to USB drives I keep in my Safety deposit box, and my working data is on a different Nas that copies to these weekly on a staggered method.

I have questions about the ZFS layout.

I am over the 10 drive Vdev (suggestion) I was considering two 8 drive ZFS2 Stripe set ~144tb.

I was considering making the 4tb AIC a write cache (is that even supported)

This data is written weekly and unless there is a failure it is only read from to copy to the USB drives for my Safety deposit box.

I am only to any and all suggestions for layout, even if I have to toss a few more bucks at it.

If a Pool cannot be used as a read/write cache I can even purchase an 8tb NVME for this use if there is a reason for it.

When it comes to a ZIL you’ll want a fast device like an optane drive. And it’s typically done on how much writes are happening in a particular time is how you size it.

If data is important then a ZIL is the way to go in case there is an outage or something happens then the data will finish committing to the drives once the system is back online.

For the raidz2, if you all about highest redundancy then I would agree with your raidz2 strategy.

So, two eight drive Vedv’s in a pool. two Truenas ZIL mini write cache one for each vdev? in Zfs2 do you think 128gb is enough or shall I max out at 256g

No real point in having a ZFS LOG that big and the LOG applies to the whole pool not just one VDEV. Memory provides ARC cache and going from 128 to 256 really depends on your cache needs. 8 wide RaidZ2 mean you can lose up to 2 drives per VDEV.

I have a video here to help you better understand caching in ZFS:

1 Like