Mutagen has two best-effort safety mechanisms aimed at avoiding unintentional synchronization root deletion or replacement. Each of these mechanisms detects an “irregular” condition during synchronization and halts the synchronization cycle until the user confirms that the condition is intentional.
In both cases, the user is required to delete the synchronization root on the
side to which the deletion or replacement should propagate, and then use
mutagen resume to continue synchronization for the session.
The first feature detects complete deletion of the synchronization root on one side of the connection. This detection is best-effort since directory deletion is non-atomic and Mutagen may see (and propagate) deletion of a large portion of a synchronization root before seeing that the entire root was deleted (though Mutagen does its best to avoid operating during concurrent file modifications when it detects them).
The second feature detects replacement of the synchronization root with a root of a different type (e.g. replacing a directory root with a file root) on one side of the connection.