Thank’s alot for all the suggestions, I found some time to make some more test and now it works!! I have tried a new backup and also made the the old backup run a couple of times with success every time!
To pin down the problem I disabled the windows 10 pro (creators fall edition) firewall on the SSH-server, not hoping for any change, but the backup completed immediately (went directly to 15 versions so the backups have worked but not marked as completed, as I suspected). Tried firewall on/off and the same behavior every time (on: finishes backup, verifies ok and then presents error message after 68 seconds instead of marking as completed).
The culprit was the windows firewall together with the automatic rules set by Bitvise SSH on that particular machine, the rules seem ok (Bitvise adds a rule that opens one TCP port when Bitvise is running and routes that to the whole subnet).
The solution was to turn off the automatic opening of the windows firewall by Bitvise SSH and add the rule manually instead (allowing both TCP/UDP if that matters). I agree it sounds strange but it is repeatable on that Bitvise SSH server. Backing up the other way the automatic opening works fine, I even double checked the Bitvise SSH settings on both machines and they were identical except the port used.
I can’t explain why the backup doesn’t get marked as finished when automatic rules are handling the firewall on only one server…this is an example of a problem not easy to pin down. Both servers running the same software’s.
One more question: if a scheduled backup fails due to server being off-line, when does it retry, only at the specified time each day/night?
Thanks!