I'm intending to create a raid1 mirror using usb3 disks attached to an
overclocked Raspberry Pi 4. It might not be going into a data center any
time soon but I've done it before and it ran without fault. It will be
used for used for general backup and media.
Last time I used btrfs only, but I want an encryption layer this time
for which I'm going to use plain-dmcrypt. My understanding is that, like
with zfs, btrfs should be proximate to the disk sectors so one can make
best use of all its features (compression, scrubbing etc). I'm
considering the following two options:
1) sda+sdb --> PV --> VG --> btrfs_pool_LV (100% of VG) -->
plain-dmcrypt
2) sda+sdb --> plain-dmcrypt --> /dev/mapper/btrfs_raid1 --> /mnt/btrfs
I look at 1) and wonder whether the lvm layers are redundant but they (I
believe transparently) allow btrfs to format sda+sdb directly .
My question is, then, in 2) once /dev/mapper/btrfs_raid1 has been
opened, is it de facto identical to /dev/sda+/dev/sdb from scenario 1)
and will therefore allow btrfs to use all its capabilities. I believe
that in scenario 1) will only have to decrypt the LV whereas in 2) I'll
have to decrypt both disks but I'm not too fussed about that.
Thanks in advance for your attention.
