3 Tor bridges specification
7 This document describes the design decisions around support for bridge
8 users, bridge relays, and bridge authorities. It acts as an overview
9 of the bridge design and deployment for developers, and it also tries
10 to point out limitations in the current design and implementation.
12 For more details on what all of these mean, look at blocking.tex in
17 Bridge relays are just like normal Tor relays except they don't publish
18 their server descriptors to the main directory authorities.
20 1.1. PublishServerDescriptor
22 To configure your relay to be a bridge relay, just add
24 PublishServerDescriptor bridge
25 to your torrc. This will cause your relay to publish its descriptor
26 to the bridge authorities rather than to the default authorities.
28 Alternatively, you can say
30 PublishServerDescriptor 0
31 which will cause your relay to not publish anywhere. This could be
32 useful for private bridges.
36 Bridge relays should use an exit policy of "reject *:*". This is
37 because they only need to relay traffic between the bridge users
38 and the rest of the Tor network, so there's no need to let people
39 exit directly from them.
41 We invented the RelayBandwidth* options for this situation: Tor clients
42 who want to allow relaying too. See proposal 111 for details. Relay
43 operators should feel free to rate-limit their relayed traffic.
45 1.3. Implementation note.
47 Vidalia 0.0.15 has turned its "Relay" settings page into a tri-state
48 "Don't relay" / "Relay for the Tor network" / "Help censored users".
50 If you click the third choice, it forces your exit policy to reject *:*.
52 If all the bridges end up on port 9001, that's not so good. On the
53 other hand, putting the bridges on a low-numbered port in the Unix
54 world requires jumping through extra hoops. The current compromise is
55 that Vidalia makes the ORPort default to 443 on Windows, and 9001 on
58 At the bottom of the relay config settings window, Vidalia displays
59 the bridge identifier to the operator (see Section 3.1) so he can pass
60 it on to bridge users.
62 2. Bridge authorities.
64 Bridge authorities are like normal v3 directory authorities, except
65 they don't create their own network-status documents or votes. So if
66 you ask a bridge authority for a network-status document or consensus,
67 they behave like a directory mirror: they give you one from one of
68 the main authorities. But if you ask the bridge authority for the
69 descriptor corresponding to a particular identity fingerprint, it will
70 happily give you the latest descriptor for that fingerprint.
72 To become a bridge authority, add these lines to your torrc:
73 AuthoritativeDirectory 1
74 BridgeAuthoritativeDir 1
76 Right now there's one bridge authority, running on the Tonga relay.
78 2.1. Exporting bridge-purpose descriptors
80 We've added a new purpose for server descriptors: the "bridge"
81 purpose. With the new router-descriptors file format that includes
82 annotations, it's easy to look through it and find the bridge-purpose
85 Currently we export the bridge descriptors from Tonga to the
86 BridgeDB server, so it can give them out according to the policies
89 2.2. Reachability/uptime testing
91 Right now the bridge authorities do active reachability testing of
92 bridges, so we know which ones to recommend for users.
94 But in the design document, we suggested that bridges should publish
95 anonymously (i.e. via Tor) to the bridge authority, so somebody watching
96 the bridge authority can't just enumerate all the bridges. But if we're
97 doing active measurement, the game is up. Perhaps we should back off on
98 this goal, or perhaps we should do our active measurement anonymously?
100 Answering this issue is scheduled for 0.2.1.x.
102 2.3. Future work: migrating to multiple bridge authorities
104 Having only one bridge authority is both a trust bottleneck (if you
105 break into one place you learn about every single bridge we've got)
106 and a robustness bottleneck (when it's down, bridge users become sad).
108 Right now if we put up a second bridge authority, all the bridges would
109 publish to it, and (assuming the code works) bridge users would query
110 a random bridge authority. This resolves the robustness bottleneck,
111 but makes the trust bottleneck even worse.
113 In 0.2.2.x and later we should think about better ways to have multiple
118 Bridge users are like ordinary Tor users except they use encrypted
119 directory connections by default, and they use bridge relays as both
120 entry guards (their first hop) and directory guards (the source of
121 all their directory information).
123 To become a bridge user, add the following line to your torrc:
126 and then add at least one "Bridge" line to your torrc based on the
129 3.1. Format of the bridge identifier.
131 The canonical format for a bridge identifier contains an IP address,
132 an ORPort, and an identity fingerprint:
133 bridge 128.31.0.34:9009 4C17 FB53 2E20 B2A8 AC19 9441 ECD2 B017 7B39 E4B1
135 However, the identity fingerprint can be left out, in which case the
136 bridge user will connect to that relay and use it as a bridge regardless
137 of what identity key it presents:
138 bridge 128.31.0.34:9009
139 This might be useful for cases where only short bridge identifiers
140 can be communicated to bridge users.
142 In a future version we may also support bridge identifiers that are
143 only a key fingerprint:
144 bridge 4C17 FB53 2E20 B2A8 AC19 9441 ECD2 B017 7B39 E4B1
145 and the bridge user can fetch the latest descriptor from the bridge
146 authority (see Section 3.4).
148 3.2. Bridges as entry guards
150 For now, bridge users add their bridge relays to their list of "entry
151 guards" (see path-spec.txt for background on entry guards). They are
152 managed by the entry guard algorithms exactly as if they were a normal
153 entry guard -- their keys and timing get cached in the "state" file,
154 etc. This means that when the Tor user starts up with "UseBridges"
155 disabled, he will skip past the bridge entries since they won't be
156 listed as up and usable in his networkstatus consensus. But to be clear,
157 the "entry_guards" list doesn't currently distinguish guards by purpose.
159 Internally, each bridge user keeps a smartlist of "bridge_info_t"
160 that reflects the "bridge" lines from his torrc along with a download
161 schedule (see Section 3.5 below). When he starts Tor, he attempts
162 to fetch a descriptor for each configured bridge (see Section 3.4
163 below). When he succeeds at getting a descriptor for one of the bridges
164 in his list, he adds it directly to the entry guard list using the
165 normal add_an_entry_guard() interface. Once a bridge descriptor has
166 been added, should_delay_dir_fetches() will stop delaying further
167 directory fetches, and the user begins to bootstrap his directory
168 information from that bridge (see Section 3.3).
170 Currently bridge users cache their bridge descriptors to the
171 "cached-descriptors" file (annotated with purpose "bridge"), but
172 they don't make any attempt to reuse descriptors they find in this
173 file. The theory is that either the bridge is available now, in which
174 case you can get a fresh descriptor, or it's not, in which case an
175 old descriptor won't do you much good.
177 We could disable writing out the bridge lines to the state file, if
178 we think this is a problem.
180 As an exception, if we get an application request when we have one
181 or more bridge descriptors but we believe none of them are running,
182 we mark them all as running again. This is similar to the exception
183 already in place to help long-idle Tor clients realize they should
184 fetch fresh directory information rather than just refuse requests.
186 3.3. Bridges as directory guards
188 In addition to using bridges as the first hop in their circuits, bridge
189 users also use them to fetch directory updates. Other than initial
190 bootstrapping to find a working bridge descriptor (see Section 3.4
191 below), all further non-anonymized directory fetches will be redirected
194 This means that bridge relays need to have cached answers for all
195 questions the bridge user might ask. This makes the upgrade path
196 tricky --- for example, if we migrate to a v4 directory design, the
197 bridge user would need to keep using v3 so long as his bridge relays
198 only knew how to answer v3 queries.
200 In a future design, for cases where the user has enough information
201 to build circuits yet the chosen bridge doesn't know how to answer a
202 given query, we might teach bridge users to make an anonymized request
203 to a more suitable directory server.
205 3.4. How bridge users get their bridge descriptor
207 Bridge users can fetch bridge descriptors in two ways: by going directly
208 to the bridge and asking for "/tor/server/authority", or by going to
209 the bridge authority and asking for "/tor/server/fp/ID". By default,
210 they will only try the direct queries. If the user sets
211 UpdateBridgesFromAuthority 1
212 in his config file, then he will try querying the bridge authority
213 first for bridges where he knows a digest (if he only knows an IP
214 address and ORPort, then his only option is a direct query).
216 If the user has at least one working bridge, then he will do further
217 queries to the bridge authority through a full three-hop Tor circuit.
218 But when bootstrapping, he will make a direct begin_dir-style connection
219 to the bridge authority.
221 As of Tor 0.2.0.10-alpha, if the user attempts to fetch a descriptor
222 from the bridge authority and it returns a 404 not found, the user
223 will automatically fall back to trying a direct query. Therefore it is
224 recommended that bridge users always set UpdateBridgesFromAuthority,
225 since at worst it will delay their fetches a little bit and notify
226 the bridge authority of the identity fingerprint (but not location)
227 of their intended bridges.
229 3.5. Bridge descriptor retry schedule
231 Bridge users try to fetch a descriptor for each bridge (using the
232 steps in Section 3.4 above) on startup. Whenever they receive a
233 bridge descriptor, they reschedule a new descriptor download for 1
236 If on the other hand it fails, they try again after 15 minutes for the
237 first attempt, after 15 minutes for the second attempt, and after 60
238 minutes for subsequent attempts.
240 In 0.2.2.x we should come up with some smarter retry schedules.
242 3.6. Implementation note.
244 Vidalia 0.1.0 has a new checkbox in its Network config window called
245 "My ISP blocks connections to the Tor network." Users who click that
246 box change their configuration to:
248 UpdateBridgesFromAuthority 1
249 and should add at least one bridge identifier.