I've also got another weird space accounting issue.
I've got a zfs dataset that is dedicated to storing a single file. A truecrypt file container that I access over cifs. According to "ls -h", the file is 900G. It is the only file on the dataset. I confirmed this with a recursive ls, and a find /mnt/zpool/dataset to list everything.
In this case though, a "du -sh /mnt/zpool/dataset" returns 2.45T. And the numbers in "zfs get all" show ~2.45T. I'm replicating this to another nas box, and what's interesting is that the size seems to be reported more correctly there. I see 1.15T on the replication target. There are snapshots of course, to support replication. But all snapshots are only using 83G.
I really have no idea how space is reported by zfs. It's way too complicated for me.
In my first example, another cifs fileshare was showing 4.4T free when the 900G zvol existed. I deleted the zvol, and went to 6T free on the other cifs shared dataset. That does actually seem like the zvol is consuming the 'wrong' space. Really weird.
I've got a zfs dataset that is dedicated to storing a single file. A truecrypt file container that I access over cifs. According to "ls -h", the file is 900G. It is the only file on the dataset. I confirmed this with a recursive ls, and a find /mnt/zpool/dataset to list everything.
In this case though, a "du -sh /mnt/zpool/dataset" returns 2.45T. And the numbers in "zfs get all" show ~2.45T. I'm replicating this to another nas box, and what's interesting is that the size seems to be reported more correctly there. I see 1.15T on the replication target. There are snapshots of course, to support replication. But all snapshots are only using 83G.
I really have no idea how space is reported by zfs. It's way too complicated for me.
In my first example, another cifs fileshare was showing 4.4T free when the 900G zvol existed. I deleted the zvol, and went to 6T free on the other cifs shared dataset. That does actually seem like the zvol is consuming the 'wrong' space. Really weird.
