This change had an unexpected impact for me. I am already running another daemon on port 8080, so when restic tried to start the rclone REST API, it failed. Prior to this change, rclone and restic communicated via a stdin/stdout pipe:
Not a big issue, I just set the RCLONE_ADDR
environment variable to unix:///tmp/rclone.sock
. However, it seems that rclone is not cleaning up the socket, because on the next restic run, I am seeing this error message:
rclone: Failed to restic: failed to init server: listen unix /tmp/rclone.sock: bind: address already in use
I have worked around this for now, by cleaning up the socket file in the bash script, that I use to manage my restic backup. I do know what the conventions for sockets are, but from my perspective it would be cleaner if rclone managed this automatically.
If this does not break any conventions, should I create a feature request over on Github?