[OpenIndiana-discuss] All of a sudden I'm feeling stupid.... Where does the COW and the newly written data go for snapshots?

[James Carlson]

On 05/12/15 09:23, Hans J. Albertsson wrote:
> I mean, I know about snapdir=visible and all that, but I want to look at
> what data has been written to a dataset after the snapshot.

The copy-on-write is done on a block basis, not a file basis.  Blocks of
data don't point back to the files in which they're contained; the
pointers go the other direction, so mapping from block back to file is
usually non-trivial.

> What I REALLY want to achieve is to see who is writing what to my
> dataset when it really shouldn't be written to!

Solution A: mount the dataset as read-only and see who complains.

Solution B: run a "diff -r" and note the changed files.

Solution C: write a small program to compare attributes using nftw.

-- 
James Carlson         42.703N 71.076W         <carlsonj at workingcode.com>