[tor-relays] relay load increase (was: note to Falback Directory operators: bandwidth wasting bot)

[Felix]

Hi everybody

Tor relay load increase is going on.

1) Not normal: Some relays see much more circuits (ie the normal number 
of circuits was about 15-20k and is 100k-400k today).

2) Not normal: Fallbacks receive extensive use today. We are writing too 
much.

Am 09.11.2018 um 17:17 schrieb starlight.2018q2 at binnacle.cx:
> For the last year a bot of some kind has loaded Fallback
> Directory relays with excessive directory document requests.
> Some detail can be found in ticket #27921 at
> https://trac.torproject.org/projects/tor/ticket/27921.
>
> To determine if your relay is affected, look the relay up in
> https://metrics.torproject.org/rs.html, click the 6-Month history
> tab, and examine the graph for significantly higher "write bytes
> per second" relative to "read bytes per second".
>
> I find this annoying (since I pay for the wasted bandwidth) and
> came up with a one-line hack that mitigates the abuse without
> impairing the usefulness of the DIR service to normal relays
> and clients.  If you compile your daemon from source and are
> interested contact me directly for the patch.

https://metrics.torproject.org/rs.html#details/
1AE039EE0B11DB79E4B4B29CBA9F752864A0259E
A2E6BB5C391CD46B38C55B4329C35304540771F1
B86137AE9681701901C6720E55C16805B46BD8E3

3) Not normal: 
https://metrics.torproject.org/hidserv-rend-relayed-cells.html?start=2017-01-01

-- 
Cheers, Felix