Any idea how many parallel download connections that used? Possibly spreading traffic helps some.
Whether or not it will make OneDrive happier, parallel connections (when possible) raise throughput.
Duplicati does them on upload because it just has to move pre-made files. Download may be harder, because files need to be worked on afterwards. One easier use might be DB recreate dblock search, although it’s still not easy. We might be heading for download (rather than SQL) limited DB recreates,
when the dindex scheme falls short, from issues that are unclear due to the usual lack of information:
Unclear if this meant wait much more than a week then hit it without a delay, or tweak the delay to
"get more throughput before throttling. I plan the latter. It’s fine overnight, but at slow 500 MB/hour.