well you can still use acd_cli but personally with all the problems with drive disconnects i would never switch back.
16 days running checking every 15 seconds if drive got disconnected and logging reconnects so far NOT 1 disconnect happend.
With acd_cli it go so retarded that i had like 50 cron script for some kind of checking and trying to remount drive.
eg tailing log to see if database got corrupted and deleting it and doing full sync.
The biggest issue I had was once my library got over 2500 folders and 20.000 files in most cases it could not sync in first itteration eg multiple syncs needed to be runned. The problem with that was when DB got corrupted the drive was disconnected and each sync took around 2 minutes before it stopped and then could not be resynced for another 5.
Acdcli worked perfectly at beginning when my lib was bellow 10TB then ...
So if rclone ever implements some sql3 db for local cashe it must be as parameter eg optional.
Atm this was my test with latest beta
3304 directories, 28933 files 32TB
14.12.2016/17:51 tree -D -s -h completed in 317 seconds
if I runned tree -D -s -h with acd_cli mount it would take less then 20 seconds as it would just read the database, but got forbid some upload and syn was being done during that time as it was more then likely DB will be corrupted.
Iam quite heavy user this is my download/upload server stats
and this is just plex
both mounted with rclone 0 problems. acd_cli really dislikes having multiple mounts.
If you dont want to risk database corruption with it then proper way is to have it mounted only on 1 machine.
When you are uploading something with acdcli disconnect the drive then sync it then reconnect it eg useless for Plex server or stuff when you want to have it on all the time, but good if you just need to access files fast, search etc...