[tor-talk] New site attempting to help Tor grow

[social at sponsor-privacy.com]

On 2015-04-17 14:55, nusenu wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
> 
>> - Bandwidth: the instances being used come with 1000TB per month.
>> This has been set as the cap in the torrc file so each instance can
>> in theory use all 1000TB before it stops passing traffic.
> 
> 1000TB?
> I guess you actually mean 1000 GB (=1TB)...?
> 
> You definitely should consider going somewhere with unmetered traffic,
> since 1TB / month isn't a lot. The relay will hit the cap fast and
> won't run for most of the month? IMHO this model doesn't scale well
> with digital ocean's pricing.
> 
> You would be able to achieve a far bigger impact if you crowd source
> bigger unmetered instances.. where you add an instance per sponsor.
> Obviously that has to be transparent to the sponsor.
> 
> 
>> - Each relay is SSD based (20gigs), comes with 1 "cpu" (these are
>> virtual instances) and 512mb of RAM
> 
> Do you have experience with running relays on 512MB of RAM?
> 
> 
>> - As far as running two relays on one instance,
> 
> given the limited available memory you probably don't want to run two
> instances
> 
> 
>> - In all honesty, the MyFamily option is something that I didn't
>> think would really apply to this as the goal is to never actually
>> login to these boxes.
> 
> You don't want to manage more than a hand full of relays without
> central management. "never actually login" is IMHO not very realistic
> long term.
> 
> -----BEGIN PGP SIGNATURE-----
> 
> iQIcBAEBCgAGBQJVMXNAAAoJEFv7XvVCELh0WI4P/RskcBzRd1ZtT6a5S52Z91lN
> nDp2CeOfdYZOWSoLoYYXG+REgjfbRoJk3yJlzdORVOPj7v70J40gf8ruiYkHsgTi
> sWbfvoSvqGoftxCqUjNMdhmcI/1DthUFPP0wvbUe72PKTNw6EdUtNnj/oEZ61zL4
> leFFePMrQ839OrKy3EUKKMYyScABT3gTAUm79utonkLpWYHsxzmP2DUGNhWTcDRu
> 7KqLPMCotlkOQPS6j1/7WqFez2z5TC2AXsZCr6lgAOpY6M883gvzLTFV8Y8lDYoc
> IyeBH7a++624DexIIjzXsviu333cy/IWwXvSP/3CqT/jCb6fwiZqDaU0ds3EjYyh
> oE2Cy8rIHdkahc5qDLh01beRql7ixSXjd9FmyU0mO2VIKC5SY1vSRQhbKyYPPG7n
> 5jcBLJqZ+tsCIv1x4/nxGNhZ1hVcu03PKwgiWA7dFTFzcapqfSB/HXuRn1ldP5R7
> eGGl5QjSZaXkQbhEagklcJO9pZyzvrqRWTd4oNsUdkIw+L87Ry4wRwjGJ99/RMuB
> KxqAtYu3kg+9rrB+HnrSanlmfMk9K3zmKwJnMe5Yf1W3f5p4QkzAbzMnRAPXginc
> MNMclV8L66zxbPnnU4+gDxOmZt9tl91uK+jpeFtvigPApfINIQa5TJdtw2XQ2CVB
> 8+QEKJUFSkszkEuDhbYc
> =roBe
> -----END PGP SIGNATURE-----

Yes, sorry, 1TB a month :)

I have been running my own personal relay on one of their 512MB 
instances for a bit over three months without issue (famous last words). 
CPU usage usually hovers around 10-15%. On that relay I set 
AccountingMax to 100 GB of traffic per day just to make sure it didn't 
blow out in a single day and then sit idle the rest of the month. I 
restarted it awhile back and looking at the logs it shows:

"Tor's uptime is 17 days 0:00 hours, with 637 circuits open. I've sent 
370.11 GB and received 369.43 GB."

Your suggestion for one big server with multiple clients on an 
un-metered account might be cool...