On Tue, Jul 05, 2016 at 10:00:22PM -0700, Green Dream wrote:
So... what's going on in this particular case and what are the directory authorities going to do, if anything?
Yesterday we started the move towards blocking them. (The move takes a little while, since it needs a sufficient fraction of directory authority operators to do it.) Specifically, it looks like 3 of the dir auths have moved to reject them, and I hear a 4th will be doing it soon, and that should be sufficient.
Speaking of which, a while ago I started a discussion of how to streamline that process: https://trac.torproject.org/projects/tor/ticket/16558 but it remains unclear whether that idea is a good one or a bad one.
As a relay operator near the top of the CW list, I continue to be somewhat uncomfortable with the lack of transparency regarding the directory authority decisions. It would be nice if the decision making process around these types of events was a bit more transparent.
First, thanks for running a relay! Second, I agree about the transparency side. Part of our challenge is that the directory authority operators, like everybody else in Tor-land, are overloaded. But that by itself is no excuse. The bigger problem is that identifying and bumping out bad relays is an inherently unbalanced situation -- unbalanced in favor of the bad relays. See https://lists.torproject.org/pipermail/tor-talk/2014-July/034219.html for more discussion on this point.
I wonder if there's a good balance we can strike, e.g. where we make it clear to the world when we decided to bump out a set of relays, since those relays are going to figure it out themselves soon enough? In this case we actually found these relays misbehaving (accessing onion addresses that they learned about), and maybe that detail is reassuring to some people, but again that arms race for noticing misbehaving HSDirs is a really crummy one from our perspective. (See also the upcoming hotpets and defcon talks by Guevara Noubir et al.)
--Roger