On Mon, 27 Apr 2015 18:18:57 +1000 Adam Goryachev <mailinglists@xxxxxxxxxxxxxxxxxxxxxx> wrote: > Speaking of which, I'm not convinced that we should spend that developer > time on each and every FS (eg, duplicated effort for btrfs, zfs, and any > others that do the same) There isn't "each and every" FS, depending on whom you ask there's just one FS that you should use. :) The filesystem can also do checksums in a smarter way, e.g. not checksum the free space. And developers of those filesystems aren't going to abandon their checksum support or plans to add that just because the underlying block device MIGHT be an MD RAID array of a new weird type. However one place where adding corruption resilience to MD can be extremely interesting and possible almost "for free" (with no disk format change, for example), is a constant full verification and corruption-healing RAID6. -- With respect, Roman
Attachment:
signature.asc
Description: PGP signature
