The error message “A task was cancelled” can mean a few things, but in this context, I am fairly sure this is a timeout guard that was introduced in 2.1.0.2 to prevent transfers from hanging forever.
The way it should work is to monitor if any data is flowing and keep resetting the timer while data flows. If there is a stall longer than 30s, this will cause the task to be cancelled.
What we have noticed is that there seems to be a pattern where this check is triggered in places where it should not, and I think your case is similar.
You can try setting --read-write-timeout=0 to disable the timeout or --read-write-timeout=5m to wait for 5 minutes before considering it a timeout.
I don’t have a good explanation for why the download would ever stall for several seconds, but I still think this is the cause of the error messages.
Let me know if it works or not, then I will dig deeper into understanding the problem.