1 Filename: 104-short-descriptors.txt
2 Title: Long and Short Router Descriptors
8 Implemented-In: 0.2.0.x
12 This document proposes moving unused-by-clients information from regular
13 router descriptors into a new "extra info" router descriptor.
17 Some of the costliest fields in the current directory protocol are ones
18 that no client actually uses. In particular, the "read-history" and
19 "write-history" fields are used only by the authorities for monitoring the
20 status of the network. If we took them out, the size of a compressed list
21 of all the routers would fall by about 60%. (No other disposable field
22 would save much more than 2%.)
24 We propose to remove these fields from descriptors, and and have them
25 uploaded as a part of a separate signed "extra info" to the authorities.
26 This document will be signed. A hash of this document will be included in
27 the regular descriptors.
29 (We considered another design, where routers would generate and upload a
30 short-form and a long-form descriptor. Only the short-form descriptor would
31 ever be used by anybody for routing. The long-form descriptor would be
32 used only for analytics and other tools. We decided against this because
33 well-behaved tools would need to download short-form descriptors too (as
34 these would be the only ones indexed), and hence get redundant info. Badly
35 behaved tools would download only long-form descriptors, and expose
36 themselves to partitioning attacks.)
38 Other disposable fields:
40 Clients don't need these fields, but removing them doesn't help bandwidth
41 enough to be worthwhile.
42 contact (save about 1%)
43 fingerprint (save about 3%)
45 We could represent these fields more succinctly, but removing them would
49 (Apparently, exit polices are highly compressible.)
51 [Does size-on-disk matter to anybody? Some clients and servers don't
52 have much disk, or have really slow disk (e.g. USB). And we don't
53 store caches compressed right now. -RD]
59 An "extra info" descriptor contains the following fields:
61 "extra-info" Nickname Fingerprint
62 Identifies what router this is an extra info descriptor for.
63 Fingerprint is encoded in hex (using upper-case letters), with
66 "published" As currently documented in dir-spec.txt. It MUST match the
67 "published" field of the descriptor published at the same time.
71 As currently documented in dir-spec.txt. Optional.
73 "router-signature" NL Signature NL
75 A signature of the PKCS1-padded hash of the entire extra info
76 document, taken from the beginning of the "extra-info" line, through
77 the newline after the "router-signature" line. An extra info
78 document is not valid unless the signature is performed with the
79 identity key whose digest matches FINGERPRINT.
81 The "extra-info" field is required and MUST appear first. The
82 router-signature field is required and MUST appear last. All others are
83 optional. As for other documents, unrecognized fields must be ignored.
87 Implementations that use "read-history" and "write-history" SHOULD
88 continue accepting router descriptors that contain them. (Prior to
89 0.2.0.x, this information was encoded in ordinary router descriptors;
90 in any case they have always been listed as opt, so they should be
93 Add these fields to router descriptors:
95 "extra-info-digest" Digest
96 "Digest" is a hex-encoded digest (using upper-case characters)
97 of the router's extra-info document, as signed in the router's
98 extra-info. (If this field is absent, no extra-info-digest
102 Present if this router is a directory cache that provides
103 extra-info documents, or an authority that handles extra-info
106 (Since implementations before 0.1.2.5-alpha required that the "opt"
107 keyword precede any unrecognized entry, these keys MUST be preceded
108 with "opt" until 0.1.2.5-alpha is obsolete.)
110 3. New communications rules
112 Servers SHOULD generate and upload one extra-info document after each
113 descriptor they generate and upload; no more, no less. Servers MUST
114 upload the new descriptor before they upload the new extra-info.
116 Authorities receiving an extra-info document SHOULD verify all of the
118 * They have a router descriptor for some server with a matching
119 nickname and identity fingerprint.
120 * That server's identity key has been used to sign the extra-info
122 * The extra-info-digest field in the router descriptor matches
123 the digest of the extra-info document.
124 * The published fields in the two documents match.
126 Authorities SHOULD drop extra-info documents that do not meet these
129 Extra-info documents MAY be uploaded as part of the same HTTP post as
130 the router descriptor, or separately. Authorities MUST accept both
133 Authorities SHOULD try to fetch extra-info documents from one another if
134 they do not have one matching the digest declared in a router
137 Caches that are running locally with a tool that needs to use extra-info
138 documents MAY download and store extra-info documents. They should do
139 so when they notice that the recommended descriptor has an
140 extra-info-digest not matching any extra-info document they currently
141 have. (Caches not running on a host that needs to use extra-info
142 documents SHOULD NOT download or cache them.)
146 http://<hostname>/tor/extra/d/...
147 http://<hostname>/tor/extra/fp/...
148 http://<hostname>/tor/extra/all[.z]
149 (As for /tor/server/ URLs: supports fetching extra-info documents
150 by their digest, by the fingerprint of their servers, or all
151 at once. When serving by fingerprint, we serve the extra-info
152 that corresponds to the descriptor we would serve by that
153 fingerprint. Only directory authorities are guaranteed to support
156 http://<hostname>/tor/extra/authority[.z]
157 (The extra-info document for this router.)
159 Extra-info documents are uploaded to the same URLs as regular
164 For extra info approach:
166 * Authorities should accept extra info, and support serving it.
167 * Routers should upload extra info once authorities accept it.
168 * Caches should support an option to download and cache it, once
169 authorities serve it.
170 * Tools should be updated to use locally cached information.
172 lefkada's exit.py script.
173 tor26's noreply script and general directory cache.
174 https://nighteffect.us/tns/ for its graphs
175 and check with or-talk for the rest, once it's time.
177 * Set a cutoff time for including bandwidth in router descriptors, so
178 that tools that use bandwidth info know that they will need to fetch
179 extra info documents.
181 * Once tools that want bandwidth info support fetching extra info:
182 * Have routers stop including bandwidth info in their router