a btrfs “array” in unraid is composed by individual btrfs partitions mounted as /mnt/disk1 /disk2 /disk3 and so on, then there’s a daemon that makes a “unified” view at /mnt/user/, and it uses a different algorithm for parity.
There’s a way to make real btrfs raid arrays, but it has been introduced very recently (1-2 years ago), and it’s not the default that you create when you use the Web UI.
The unraid array it’s not an array, but a bunch of xfs partitions where a third party program is doing softraid. If opened on any other Linux distro, they will mount but parity drives are ignored
If you have new drives: make a zfs array and copy all files there
If you want to recycle drives while temporarily keeping the parity drives: from unraid 7 set a drive as unused, use the mover (or unbalance) to empty it, check if it’s actually empty by going to /mnt/diskX , format it as btrfs, set it as preferred for your shares, choose another disk to empty, use the mover to move all the data from the next disk to the new btrfs one, then remove the empty drive and add it to the btrfs raid. Repeat.
for some reason i read xfs instead of btrfs…
their array implementation is stupid and smart at the same time. I love that you can mix and match disks at random, even with different filesystems (you can have one drive in murderfs, one in xfs, one in btrfs and the last one in zfs and all protected with the same parity drive), but i completely hate how my server is locked by a 25% iowait due to how much cpu intensive is their softraid. Maybe when they came out with this system 20 years ago it was groundbreaking, but now it is stupid. Now with unraid 7 released one week ago they’re starting to deprecate it (disabled by default in new install) in favor of native btrfs or zfs arrays