> I don't think btrfs has a concept of having only some subvolumes usable. Either you can mount the filesystem or you can't.

You can still mount the BTRFS array as degraded if you specify it during mount. But then this lead to some others issues like the missing data written while degraded will not be automatically be copied over without doing a scrub, while ZFS will resilver it automatically, etc

> You can prevent the issue in the first place by identifying the filesystem by UUID instead of by an individual block device.

I tried that, but all it does is select the first available block device during mount, so if that device goes down, the mount also goes down.