[network-health] Upgrade your 'eax' Tor relay?

[Roger Dingledine]

On Sat, Oct 05, 2019 at 06:55:04PM -0400, Ehsan Akhgari wrote:
> I upgraded my node to the latest version, so hopefully that issue should be
> resolved now.  Thanks for your ping!

Great. It looks good now.

> About the bandwidth, I checked the tor wiki docs and also my linode
> instance limits again, and I think I could drop the bandwidth rate limits,
> and just keep a total monthly cap to ensure my account doesn't exceed its
> monthly quota.  Based on my understanding of the documentation this should
> be a better setup, is that correct?
> 
> Here is what my torrc looks like right now:
> 
> AccountingMax 3500 GB
> AccountingStart month 1 00:00

Yes, that will work. But two potential catches to be aware of:

(A) That's 3500 gigabytes in each direction, and some ISPs count a byte
in and a byte out as two bytes for your monthly cap.

(B) If I'm doing my math right, 3500 GBytes per month is 1.35 MBytes/s
on average. You might watch take a look at it once or twice a month,
e.g. on its relay-search page, and see how much of the month it spends
hibernating. If it's a lot, you might put in a few additional rules, like
RelayBandwidthRate 3 MBytes
RelayBandwidthBurst 50 MBytes
to limit it to a long-term average of 3MB/s while still giving it a lot
of room in the token bucket before the rate limiting kicks in.

See also this original faq entry:
https://2019.www.torproject.org/docs/faq#LimitTotalBandwidth

Thanks!
--Roger