On Sat, 7 Jan 2012, 14:27:19 EST, Craig Sanders <cas@taz.net.au> wrote:
URL?
Umm, not sure, it's in my inbox. :-) This looks like the thread: http://comments.gmane.org/gmane.comp.file-systems.btrfs/15027
OK, that sounds like it's some sort of optimisation in ext4 so that subsequent fstrim runs are faster.
Correct.
I'm still not sure why fstrim says (on the 2nd and subsequent runs on XFS) that it has trimmed almost the same amount of data.
Because XFS doesn't remember what was trimmed so sends the same list (plus any new ones) down to the FS I'd guess.
what is it actually counting here? bytes in data blocks that (might) need trimming? byte-counts in TRIM commands sent to the device? something else?
Not a clue I'm afraid - this man page indicates its bytes the FS has sent to the block layer to be trimmed: http://manpages.ubuntu.com/manpages/oneiric/man8/fstrim.8.html cheers, Chris -- Chris Samuel - http://www.csamuel.org/ - on mobile