On SCALE Electric Eel 24.10.2, I have an HDD pool that contains multiple datasets. I have a separate SSD pool that contains a Linux Mint VM (specifically Mint version 22.1).
The datasets on the HDD pool are shared via SMB, and when I use the Mint VM to copy files between datasets on the HDD pool, they transfer almost instantly.
For example, today, from within the Mint VM, I copied a 135GB file from one dataset on the HDD pool to another dataset on the HDD pool, and the transfer took less than 1 minute. (I could understand if it were a move, but it was a copy.) This speed is clearly impossible on an HDD pool unless there is something going on behind the scenes that doesnât physically move all the blocks of data.
Also, it seems to me that these lightning fast copies only started happening recently, with recent versions of SCALE and recent versions of Linux Mint as the client used to do the transfer over SMB.
Can anyone figure out how and why this is happening? Obviously, I donât mind fast transfer of data between datasets, but I would like to figure out whatâs going on so that I donât run into any problems.
Thanks. That sounds right. Whatâs interesting is that SMB and Linux Mint are smart enough so that block cloning works on exposed datasets, especially since the github discussion you linked to says âmost versions of Linux will not allow clones across datasets,â and in my case, this behavior is happening with file copies between different datasets although within the same pool.
It canât be SMB server side copying. Itâs way too fast for that. 135 GB transferred in less than one minute would be more than 2 GB per second transfer speed, which is impossible on a hard drive.
Unless someone has other ideas, I believe it almost has to be block cloning. Perhaps the restriction on transfer between datasets has been lifted.
