Syncthing has worked for a decade on all kinds of networks, including today. I have one very unusual connection: highly assymmetric 250-350 Mbps down, 10-20 Mbps up. It is high latency (33ms unloaded / 470ms loaded according to fast.com). The other side is on a fibre 1/1 Gbps, rock solid and stable in any other case. Connection defaults to TCP WAN 1+2 IPv4 (before restart just now, it was IPv6)
I have noted this with Syncthing and other applications, if upload is hogged, download barely works, probably due to lacking bandwith for control messages. If no big uploads are queued (<10 MiB), downloads are speedy, sometimes over 400 Mbps. Yet when there is a 100-200 MiB upload taking multiple days, the downloads are hogged.
I tried limiting upload speed (and download speed on other side, A, B, A+B), and manually setting connection count, without results. Is there some other method to prioritize control messages by throttling the (data) upload bandwidth?
https://www.waveform.com/tools/bufferbloat?test-id=c112871c-0994-47be-b202-b0b131542cc4
I’m more interested than motivated, as the problematic connection will cease use in a month or two.