Hi,
On Tue, Aug 05, 2025 at 08:49:43AM +0200, fred.kite.dml@mailo.com wrote:
Would it be safe to perform the deduplication with a single command on
/home instead of each user's folder separately? Will it create
problems if the same file is found in several home folders but has
different ownership and permissions?
It can leak information about what is shared, but this is perhaps not of
concern in your case.
For example, if user "foo" has a large file and runs "filefrag -e" on it
then they can see the filesystem extents that compose the file. Your
system later deduplicates this file by making one or more of its extents
point to the same extent(s) within the storage of user "bar". User "foo"
can now run "filefrag -e" again and see that some extents have changed.
They now know that this data *also* exists somewhere else on the
filesystem. They don't know where, but this is not information that they
had before.
Secondly, and this is not related to cross-user deduplication but more a
question of the amount of deduplication: on HDD media frequent
deduplication can fragment files into lots of pointers to duplicate
extents, reducing streaming file read performance as a sequential read
has to move the read about across the disk. Obviously not much of a
factor for non-rotational media.
That's all that comes to mind.
Thanks,
Andy