I just signed up for the annual subscription, but I’m unable to connect to any cloud service.
Whatever I choose, I get the following error message:
Config failed to refresh token: failed to start auth webserver:listen tcp 127.0.0.1:53682 bind:address already in use
My machine is running Windows 11, and rclone, which is installed on a server on my LAN, appears to be connected correctly.
I’ve already confirmed that there are no other services listening on that port.
As you can see from the screenshot, there is nothing listening on port 53682.
First of all, apologies for the late reply and thank you for your patience.
The error you’re seeing —
Config failed to refresh token: failed to start auth webserver: listen tcp 127.0.0.1:53682 bind: address already in use
usually occurs when rclone tries to start its temporary local webserver during the authentication process. Even if nothing else appears to be listening on that port, Windows may sometimes reserve or block certain ephemeral ports, preventing rclone from binding correctly.
A couple of possible causes are:
Another background service briefly using the same port.
A firewall or antivirus program interfering with local loopback connections.
Multiple instances of rclone running simultaneously and trying to use the same port.
If the problem persists, I recommend posting the details (including your rclone version and logs) on the official community channels, where the rclone developers and advanced users can provide more targeted help:
I know this may sound crazy, but this just happened to me yesterday after trying to get a few additional settings added to the global settings to find out I was adding them wrong, and then some of them just would not work the same as I expected them to (you have to have each setting separated with a ; and no spaces … ie. —fast-list;—vps….
when that field was messed up, I got the same error where it would not start, even tried to update rclone. When I reset everything to default (on the settings tab I believe) it worked again.
Yes, that’s correct. We found a bug in how global flags are processed when they include both an option and an additional parameter, such as --transfer “30”. Please refer to the following topic for more details. Thanks for your feedback!
