Synchronising over SSH

Source or target may be on another computer that can be reached over SSH – a NAS, a server or a small computer in the home network, for example. Both sides at once are possible as well.

Setting up a connection

File → New SSH connection… or Edit → SSH connections… opens the management. A connection consists of:

Name	Free to choose. Paths later refer to the connection by this name.
Host	Computer name or IP address.
User	Account on the remote computer.
Password	Stored encrypted in the configuration, not in plain text.

In the overview, connections can be created, edited, renamed and deleted. If a connection is renamed, the synchronisations that use it follow automatically.

Writing remote paths

A remote path consists of the name of the connection, a colon and the path on the remote computer:

nas:/home/anna/documents

In the dialog of a synchronisation, the SSH button next to source and target first leads to the choice of the connection and then to a dialog that browses the directories on the remote computer – with size and modification date, so that the right place can be found without typing.

What is different

Symbolic links on the remote computer are not followed. They are skipped, so that a link to a parent directory cannot cause an endless loop.
Deleting is permanent. The recycle bin is only available locally; a remote file would first have to be downloaded to be moved there. This concerns Also delete deleted files in the target as well as Delete the source after copying, although the latter is not carried out for remote sources anyway.
File attributes are not changed on the remote side.
If the server cannot be reached, AutoFileSync tells this apart from “there is nothing there any more” and leaves the target alone. A failed connection therefore never leads to deletions.
On the first connection the server's key is stored. If it changes later, AutoFileSync aborts the connection and reports a suspected man-in-the-middle attack.

The check before every run

Before a synchronisation with a remote side starts at all, AutoFileSync makes sure that the other end is there: it connects, checks the server's key and logs in. This is the same question that “is the drive connected?” answers for a local path. If one of these steps fails, the synchronisation is skipped for this run – nothing is copied or deleted.

Source and target are held to different standards:

The source must exist. If the server answers but reports “does not exist”, the source has disappeared – and a synchronisation whose source has disappeared keeps still instead of repeating the disappearance. Otherwise a deleted source directory together with Also delete deleted files in the target would empty the target.
For the target it is enough that the connection succeeds. A target directory that does not exist yet is created by AutoFileSync itself – it must not block the synchronisation.

The result is cached, so that not every run opens a new connection. A reachable remote side is checked again at most once an hour, an unreachable one at most every five minutes. After a fault it can therefore take a few minutes until the synchronisation starts again by itself.