@buzzbuzz Welcome to the Rockstor community.
Yes this is a know limitation on the client side when using gvfs (which in turn uses libsmbclient) where as if you test the same hardware having made the mount via for example 'mount -t cifs …" you would see the expected full line speed. Rather unfortunate that the linux desktop clients don’t have this one sorted yet. But it use to be much slower.
You should be able to test the difference by first mounting, as I suspect you already have, via libsmbclient/gvfs ie via most gui tools such as nautilus or whatever the kde file manager is called these days “smb://rockhp.lan/samba-test/” style. And then compare those results with a mount achieve via for example:
sudo mount -t cifs -o username=test,password=testpass //rock.lan/samba-test /mnt/
which bypasses libsmbclient.
My tests a while while back gave 102 to 104 MB/s for the straight mount t cifs but only 38-58 MB/s for the nautilus / smbget type arrangement.
One can also do a gvfs mount via the command line with something akin to:
gvfs-mount smb://SERVER-IP/Share
to rule out other elements.
It’s been a very long standing issue that is quite disappointing for us linux desktop users.
This has come up before and you might find the discussion of it in the following closed pull request of interest:
EDIT OK forgot to post the relevant link so here it is:
ie 103 MB/s vs 48 MB/s for the 2 different mount methods.
See particularly my comments on 26 Nov 2016 as it contains upstream references acknowledging the performance problem. I haven’t looked into this again since that post but I recently only saw line half speed on my fedora desktop after mounting via Nautilus.
So essentially this is a linux desktop client issue not a Rockstor samba server issue.
Hope that helps.