Hi There I am actually a little bit confused now. I am one of (as Rana knows) those Pi 3 based OR operators with daily changing IP address. My consensus weight is about 5,000, with a max. of 1,400 connections. I would like to recomment, that the TOR org should publish minimum requirements to run a relay. And, if my kind of relays would bring disadvantage to the network, I would shut it off. So far, I think, that small and large relays in combination are good for anonymity. Mike
@balbea16
I am actually a little bit confused now. I am one of (as Rana knows) those Pi 3 based OR operators with daily changing IP address. My consensus weight is about 5,000, with a max. of 1,400 connections. I would like to recomment, that the TOR org should publish minimum requirements to run a relay. And, if my kind of relays would bring disadvantage to the network, I would shut it off. So far, I think, that small and large relays in combination are good for anonymity.
I am even more confused than you. My 1300 connections relay has a consensus weight https://atlas.torproject.org/#details/707A9A3358E0D8653089AF32A097570A96400CC6 of 38 (thirty eight). That’s less than 1% of your weight, despite having 26% the number of connections you have. Besides, I could never understand why people measure the “size” of the relay by the number of connections. My guess is you can have a large number of dead connections.
This is just another example of the lack of clear metrics and feedback to relay operators on the quality of their relay, including how well the relay is doing in terms of its usefulness to Tor, and especially of (direct or easily derived) actionable recommendations on how to improve its quality, including abandonment if it is harmful to Tor in its current form.
Rana
PS I guess this may not be that simple as possibly some relays are getting bashed because they are suspected of being rogue. Still, a mechanism of quality feedback is needed for the (hopefully the majority) of the relays that are run by bona fide volunteers.
@balbea16
Correction: your relay has 1400 connections, mine has 1300. Therefore my relay has less than 1% of the weight of your relay while having ALMOST THE SAME number of connections. Go figure…
From: tor-relays [mailto:tor-relays-bounces@lists.torproject.org] On Behalf Of Rana Sent: Saturday, December 24, 2016 10:12 AM To: tor-relays@lists.torproject.org Subject: Re: [tor-relays] What's a "useful" relay?
@balbea16
I am actually a little bit confused now. I am one of (as Rana knows) those Pi 3 based OR operators with daily changing IP address. My consensus weight is about 5,000, with a max. of 1,400 connections. I would like to recomment, that the TOR org should publish minimum requirements to run a relay. And, if my kind of relays would bring disadvantage to the network, I would shut it off. So far, I think, that small and large relays in combination are good for anonymity.
I am even more confused than you. My 1300 connections relay has a consensus weight https://atlas.torproject.org/#details/707A9A3358E0D8653089AF32A097570A96400CC6 of 38 (thirty eight). That’s less than 1% of your weight, despite having 26% the number of connections you have. Besides, I could never understand why people measure the “size” of the relay by the number of connections. My guess is you can have a large number of dead connections.
This is just another example of the lack of clear metrics and feedback to relay operators on the quality of their relay, including how well the relay is doing in terms of its usefulness to Tor, and especially of (direct or easily derived) actionable recommendations on how to improve its quality, including abandonment if it is harmful to Tor in its current form.
Rana
PS I guess this may not be that simple as possibly some relays are getting bashed because they are suspected of being rogue. Still, a mechanism of quality feedback is needed for the (hopefully the majority) of the relays that are run by bona fide volunteers.
On Sat, 24 Dec 2016 10:11:51 +0000, Rana wrote:
@balbea16
...
I am even more confused than you. My 1300 connections relay has a consensus weight https://atlas.torproject.org/#details/707A9A3358E0D8653089AF32A097570A96400CC6 of 38 (thirty eight). That???s less than 1% of your weight, despite having 26% the number of connections you have.
Number of connections is a very indirect metric. Especially any somewhat-used relay will have connections open to most of the bigger relays, since the paths for which this relay is selected come from and go to arbitrary relays, weighted by their bandwidth. So if you have few connection, this is an indication that your relay doesn't get used much. (My numbers are around 4000 and 2000 for the two of mine, with a factor of eight between them in the carried traffic).
Besides, I could never understand why people measure the ???size??? of the relay by the number of connections.
I never noticed anybody did. The number of connections only come into play when
- your VPS has a connection limit that is too low,
- you're connected to the internet via a NAT/state-keeping router/modem/thingy that can't handle as many simultaneous connections.
My guess is you can have a large number of dead connections.
Actually no. They time out. And probably actually yes, as many connections may not be in use at a specific given point.
Andreas
tor-relays@lists.torproject.org
-
Andreas Krey -
balbea16 -
Rana