Ok.. its more an issue of not really knowing how ZFS handles this and after some reseach still confused
I am going to round off numbers and generalize for simplicity
in the past using ZFS with OS X server, lets say I had a pool with 3TB space and 3 filesystems that mounted as 'disks' that osx presented as volumes
without reservations or quotas.. OSX server would show 9tb of space available (proveded the volumes were empty) because it would see that there was the entire size of the pool availabe in each child or volume...
so I set a reservation and quota of 1TB for each dataset pool/diska, pool/diskb, etc
that showed each disk that was mounted now as a 1TB volume with get info or OSX server and as space was used on each disk, it would report space remaining correctly
now using 1.4.5 I noticed that the pool let say is 10.9TB using list
if I do a zfs list it shows the root dataset as 10.7tb not 10.9?
if I reduce the reservation and quota on any of the datasets it does not give back to the root dataset.. ie if I had in the above example set each to 1TB the root dataset should show that all 3 TB are used and that 0tb available, but if I reduce the reservation and quota on say pool/diska .. then pool should show 2.5TB used with .5tb avail and pool/diska as .5TB available and say 0 used.. ?
so what am I doing wrong here.
I dont use the root dataset although it mounts.
I want the data sets that mount to have defined sizes so that finder shows a size, used and available that is a correct slice of the underlying pool without reportd 'duplicated' free space
thanks guys