Right now I’m living in my brother’s condo and he has Frontier FiberOptic 500 Mbps. On that connection I am running four middle relays: Relay Search
One problem is in the past week or two, a middle relay is using 100% of the allocated bandwidth. I had my mom and brother complain about slow internet so I had to cap my relay speeds to 320 Mbps total (80 Mbps per relay) to give enough buffer space for non-Tor traffic.
It’s interesting how the tides have turned, I complained relays don’t push enough traffic and now they’re pushing too much. Is there a denial of service attack on the Tor network, say attacking onion services? My exit relays aren’t at 100% capacity despite a funky exit policy allowing BitTorrent traffic but being “reduced” under port 10000.
I looked at your relays just now. Seems like their consensus weight was 0 from March to June. That’s the reason they weren’t getting traffic. The question is why the weight was 0.
I was in the process of moving from Seattle so my relays were off during moving. That’s why I had a 0 consensus weight.
If you stay offline for an extended period of time, the consensus weight resets to 0. My family’s considering a NYC home without fiber so it’ll go offline again until I get Verizon to expand fiber (fiber on demand is part of NYC law).
Maybe your question is more about why this happened rather than what to do about it… But about network congestion, if you have access to the router is there a QoS setting of some kind that will let you prioritize personal traffic over relay traffic?
I have an update: it seems the DoS attack (or whatever happened) is gone. My relay bandwidth use is back at “normal” levels.
I do hope Arti gets relay support sooner than later, since my server’s Xeon 5412U cores aren’t exactly all that powerful as server chips focus on maximizing core count. But 24 cores could easily make a ~10 Gigabit Tor relay.