On 19.02.19 г. 9:33 ч., Peter Becker wrote: > I would prefer "< 2" .. in both cases (RAID1 and DUP) because this > allow us to implement N-Copy RAID1 in the future. NAK When/if N-Copy raid1 patches land then they can modify the code and it will be visible from the commit message why it's < 2. For now we only support ordinary raid1, hence the code should do just enough to support the current design. Even now there are artifacts of similar "future proofing" for patches that never landed i.e dedupe stuff and pointless args to extent_clear_unlock_delalloc and some other functions.
