- tor-dev - lists.torproject.org

Summary of meek's costs, May 2015
by David Fifield 04 Jun '15

04 Jun '15

Here's the summary of meek's CDN fees for May 2015. App Engine + Amazon + Azure = total by month February 2014 $0.09 + -- + -- = $0.09 March 2014 $0.00 + -- + -- = $0.00 April 2014 $0.73 + -- + -- = $0.73 May 2014 $0.69 + -- + -- = $0.69 June 2014 $0.65 + -- + -- = $0.65 July 2014 $0.56 + $0.00 + -- = $0.56 August 2014 $1.56 + $3.10 + -- = $4.66 September 2014 $4.02 + $4.59 + $0.00 = $8.61 October 2014 $40.85 + $130.29 + $0.00 = $171.14 November 2014 $224.67 + $362.60 + $0.00 = $587.27 December 2014 $326.81 + $417.31 + $0.00 = $744.12 January 2015 $464.37 + $669.02 + $0.00 = $1133.39 February 2015 $650.53 + $604.83 + $0.00 = $1255.36 March 2015 $690.29 + $815.68 + $0.00 = $1505.97 April 2015 $886.43 + $785.37 + $0.00 = $1671.80 May 2015 $871.64 + $896.39 + $0.00 = $1768.03 -- total by CDN $4163.89 + $4689.18 + $0.00 = $8853.07 grand total The number of users stayed in the range of 3000 to 4000. There are a few days of missing data. I think the big increase to 6000+ users on the day before the gap is a bad measurement. https://metrics.torproject.org/userstats-bridge-transport.html?graph=userst… On May 22 I asked the operator of the meek-google bridge to rate-limit the bridge to 1.5 MB/s in an attempt to limit costs. Since then we further limited it to 1.1 MB/s. It pains me to do that, but I need to find a way to make the operation of this transport sustainable. I estimated the rates to keep us under the $500/month credit we have from Google. I may also rate-limit meek-amazon too. If you want to set up your own bridge and CDN instance without rate limiting, I can help you do that if you email me. Here are some docs to look at: https://trac.torproject.org/projects/tor/wiki/doc/meek#Howtorunameek-server… https://trac.torproject.org/projects/tor/wiki/doc/meek#GoogleAppEngine https://trac.torproject.org/projects/tor/wiki/doc/meek#AmazonCloudFront https://trac.torproject.org/projects/tor/wiki/doc/meek#MicrosoftAzure == App Engine a.k.a. meek-google == Total bandwidth was down, probably because of the new rate limiting. Oddly, instance hours were way up, perhaps for the same reason. The rate limiting happens at the bridge, so it's possible that it's wasting resources at the App Engine level. Here is how the Google costs broke down: 5359 GB $643.07 4571 instance hours $228.57 Compared to the previous month: 6304 GB $756.59 2597 instance hours $129.84 In the graph you can really see the impact of limiting bandwidth. Interestingly, the number of users did not decrease. It's just going slower for them :( https://globe.torproject.org/#/bridge/88F745840F47CE0C6A4FE61D827950B06F9E4… == Amazon a.k.a. meek-amazon == Requests were up about 13% and bandwidth was up about 15%. Asia Pacific (Singapore) 132M requests $158.86 973 GB $129.18 Asia Pacific (Sydney) 633K requests $0.79 4 GB $0.58 Asia Pacific (Tokyo) 47M requests $57.04 331 GB $43.14 EU (Ireland) 184M requests $221.61 1847 GB $147.56 South America (Sao Paulo) 7M requests $17.04 49 GB $11.73 US East (Northern Virginia) 68M requests $68.57 509 GB $40.27 -- total 441M requests $523.91 3715 GB $372.46 https://globe.torproject.org/#/bridge/3FD131B74D9A96190B1EE5D31E91757FADA1A… == Azure a.k.a. meek-azure == meek-azure increased about 30% again this month. You should only count on it working for another 3 months. https://onionoo.torproject.org/bandwidth?fingerprint=AA033EEB61601B2B7312D8… Estimated bandwidth use this month: 2015-05 2592 GB Compared to last month: 2015-04 1982 GB https://globe.torproject.org/#/bridge/AA033EEB61601B2B7312D89B62AAA23DC3ED8… David Fifield Earlier reports in this series: https://lists.torproject.org/pipermail/tor-dev/2014-August/007429.html https://lists.torproject.org/pipermail/tor-dev/2014-October/007576.html https://lists.torproject.org/pipermail/tor-dev/2014-November/007716.html https://lists.torproject.org/pipermail/tor-dev/2014-December/007916.html https://lists.torproject.org/pipermail/tor-dev/2015-January/008082.html https://lists.torproject.org/pipermail/tor-dev/2015-February/008235.html https://lists.torproject.org/pipermail/tor-dev/2015-March/008427.html https://lists.torproject.org/pipermail/tor-dev/2015-April/008596.html https://lists.torproject.org/pipermail/tor-dev/2015-May/008767.html

1 0

Researching Tor: Quantifying anonymity against a global passive adversary
by Florian Rüchel 03 Jun '15

03 Jun '15

Hi there, I am currently writing my Master's thesis about Mix networks & Tor (was on list previously). I am currently at a point, where I'd like to practially quantify anonymity. That is, given a pcap, I want to anaylze how successful an adversary can determine whether a specific client talked to a specific service. Since I run all of this in a simulation (using Shadow), I have access to ground truth and can determine the traffic, network size, etc. arbitrarily. The idea behind all of this is to integrate a Mix relay on each Tor node and to watch how the adversarys success behaves. However, I have a hard time finding material on such attacks, i.e. papers that closely examine this scenario (and perhaps have conducted some simulation and/or measurements themselves, beyond simple theory). The thing is that this is only part of the research and I'd really prefer if I could build this upon work of others instead of starting from scratch. Do any of you know of such papers or corresponding research? Regards, Florian

3 2

onionoo: bug in family set detection?
by nusenu 02 Jun '15

02 Jun '15

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 Hi, by comparing different methodologies of "parsing" myfamily data I stumbled upon differences between onionoo and compass. After manual review I assume there is a bug in onionoo (or onionoo has a different opinion on what families actually are) Example: According to onionoo, torpidsDEevanzo [1] is part of a family with 38 members. It lists torpidsFRonline [2] as one of its members, that implies that torpidsFRonline lists torpidsDEevanzo as one of its members as well, but torpidsDEevanzo does _not_ list torpidsFRonline (according to onionoo data). grep 'fingerprint":"0C77421C890D16B6D201283A2' details.json|grep 5510FC1736B16D46D3F2DDA5011995C478D42594 (no result) Is this a bug? thanks [1] https://atlas.torproject.org/#details/5510FC1736B16D46D3F2DDA5011995C478D42… [2] https://atlas.torproject.org/#details/0C77421C890D16B6D201283A2244F43DF5BC8… -----BEGIN PGP SIGNATURE----- iQIcBAEBCgAGBQJVbLCAAAoJEFv7XvVCELh0bxkP/1Ep/fhLMhHGIgYnVzEj+xyJ Sb4gqUBuwLckcXLAYGoKZkIeHNS4TFLBidKxuP54bvSLgqPgpsy+OzLiaUk8AYyP Ea2tTmDqcjWy2v/cpP43M3Vh+TpPBTMenNajKwe9Moz63pMpo0V9HLnBjCwBN+Mh 1vGFJw/ov4wmQO3J5OOEABdVt8RTmqCWJBVbz1NmKWNEpxa5sbPS+6tNEvBLbsaL XmtoPcdyAl+14QQk1GSq4Px49K4MFEcRuCN2r1bzRv2WaPgRmFUtrK9JCDwwLlV5 aSrBU7S2M5gn6KowkEZllniZ7n5Ze3SiifqJEPm3hyIUXczveR02LHk++lEXneQE pqxDX6PPnVIt7lm1JW5Svr82boYTC1LQ9xHyF9LStBfYvgWSD5TrDeNaIe8+mka3 OrJ6r0hxMB1IRUaxouI/IGMZLmwj2lqifBeaNccsz876oJ2W6SZu6JPVvNstf7lc TnesBplpAprekofUb3DaGeM80Pa/NmuRjga5F+H6HLH3qE8AxeGKzSQDKYLuRlaM GdZY/bMN/6JGhoHCvYIT8NMNGMYfMPE7UWiHtX+1dezIwyYtwnHxiAcRl/e73TuI cmtzvIiKA9Z1sM2FX8mhaRi6r6TOBdvovSTkKtxkXHk2qT1Sf/oPtvlGtzTlrVp7 uFFgl0Sg++eiHAd1VG58 =D6Qh -----END PGP SIGNATURE-----

2 2

Re: [tor-dev] onionoo: bug in family set detection?
by teor 02 Jun '15

02 Jun '15

Date: Tue, 02 Jun 2015 16:52:00 +0000 From: nusenu <nusenu(a)openmailbox.org> > > teor: > > MyFamily requires bidirectional declarations to be effective. > > I'm aware of that fact ;) > > > In this case: OnionOO appears to correctly implement the > > bidirectional MyFamily logic > > Apparently it doesn't. > > Since onionoo says there is a bidirectional "connection" (family) between > 5510FC1736B16D46D3F2DDA5011995C478D42594 => > 0C77421C890D16B6D201283A2244F43DF5BC89DD > > but there is none.. (as explained in my last email) > > Compass does *not* say that there is a bidirectional connection > between those relays.. onionoo says there is one even though I can't > see it, do you see it? I'm sorry, I must have got the onionoo and Compass results mixed up somehow. In Compass, I see that: 5510FC1736B16D46D3F2DDA5011995C478D42594 has no family. In Globe/Atlas (based on onionoo), I see that: 5510FC1736B16D46D3F2DDA5011995C478D42594 has a very large family. So my criticism of Compass actually applies to onionoo. teor teor2345 at gmail dot com pgp 0xABFED1AC https://gist.github.com/teor2345/d033b8ce0a99adbc89c5 teor at blah dot im OTR D5BE4EC2 255D7585 F3874930 DB130265 7C9EBBC7

1 0

OS X OpenSSL/Tor Build Instructions
by teor 02 Jun '15

02 Jun '15

As noted recently, OS X comes with OpenSSL 0.9.8 by default, which is too old to build Tor. During the Tor patch workshop today, rl1987 and Yawning were testing the OpenSSL and Tor builds on OS X. I've put the OS X build steps they used on the Tor Wiki, along with the alternative method of obtaining OpenSSL and/or Tor via MacPorts: https://trac.torproject.org/projects/tor/wiki/doc/MacBuild If you're building Tor on OS X, please try these instructions, fix any mistakes, and add missing steps or other alternatives. In particular, HomeBrew instructions for OpenSSL and Tor would be really helpful for some users. teor teor2345 at gmail dot com pgp 0xABFED1AC https://gist.github.com/teor2345/d033b8ce0a99adbc89c5 teor at blah dot im OTR D5BE4EC2 255D7585 F3874930 DB130265 7C9EBBC7

1 0

Re: [tor-dev] The Onion Name System (OnioNS)
by OnioNS Dev 02 Jun '15

02 Jun '15

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 > What's to stop a sybil attack where the malicious relays try to occupy likely site(s) for the next Quorum? > > Is the consensus unpredictable enough to thwart this attack? > Even during quiet times? (Does Tor have quiet times?) As you can see from the ACM paper, I estimated 16Kb - 28Kb from routers leaving/joining the consensus, and my Markov Model analysis shows ~9000 bits from routers changing in some degree or another. Certainly the size and dynamics of the Tor network are enough to put the entropy above 256 bits, so I'm confident that any sort of malicious maneuvers (such as changing router descriptors so that the SHA-384 hash results in an attacker-pleasing Quorum) is hard because of the cryptographic defenses of SHA-384 and because the consensus has a large degree of entropy. I believe (but cannot prove) that those numbers are a lower-bound, though formally it's just an estimation. Sybil attacks are possible but expensive. They would have to gain the Fast and Stable flags and be subtle enough to not be detected by the Tor community (the Lizard Squad learned that the hard way). Sybil attacks would also likely compromise Tor and hidden services, which is outside my security assumptions. > Can we ensure the Quorum servers have to be long-lived and high-capacity (for example, guards) to make it harder to spin up servers and immediately be added to the Quorum? > I'm not convinced the Stable flag is hard enough to get. Of course, there's a trade-off where making the set of Quorum Candidates too small makes the Quorum easier to predict, too. One possible counter-measure is the Sybil attack detector that was posted here the other day. It might be a good idea, I'll think more about that. > By the way, in your ACM paper 5.4.2 you switch from Charlie to Alice, but I think they're meant to be the same Quorum Candidate. Thanks, I'll take another look. Alice is always a Tor client. > The HS is down, and has been for some days. Yes. I'm currently hosting it myself and I'm currently but temporarily in a remote location where the Internet is spotty at best. By the time I deploy OnioNS into beta or into production I'll have a stable host for the HS. I'll have it up soon. In the meantime, you can always clone and browse the Github repo that powers it, OnioNS-www. > I have two questions about the Stem-based approach: > 1. Applications which use Tor via SOCKS won't be able to use .tor domains without using Stem. This adds another dependency to apps which want to use .tor addresses, and confuses users by making .tor addresses work with some torified apps, and not others. Is your final goal to have .tor lookups contained within the Tor client? I'm sure someone could help with the C coding if that's the issue, particularly given a working implementation in Python/Stem. If a user wants to use a .tor address with something other than the Tor Browser, they will have to have both the Stem and the OnioNS client software running in the background. It should be application-independent. Eventually I would like the Stem code in the Tor client, and theoretically it should not be hard to do so. Currently I have a very reliable Stem solution (for which I am very thankful) and I will be utilizing that script for the time being. Based on a quick investigation it should also be easy to auto-launch it with the Tor Browser. I would like to see the OnioNS package in Linux or Tor repositories so that it may work as a general solution, and not something tied to the Tor Browser. > 2. I always feel nervous when people say "letting app X spin". I assume Tor Browser just looks like it's doing a DNS lookup or similar? I may have misused the term. It's not a busy-wait. Currently it's an idle synchronous wait on separate thread in the Stem script. The Tor Browser appears as if it's waiting for a DNS lookup to occur, which is essentially correct. Once that goes through, Tor attaches the stream to a .onion address, and the browser transitions to appearing as if it's waiting for a server over a high-latency connection, which is also correct. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQEcBAEBCAAGBQJVbg1lAAoJEK2XNk/CC+yAnxEIAMn+XvrKX02+1OGrxQe4poNu hJzDcZHIJ5GTkYJk6iIOu7J+riz/f7rmQbvtxJWt3ZcAfebqcoec0j1kbvnFWDP4 /C35A4nSEN36XlDU/5AkYHv6l5dgvvxX+cDMSvHKFzrZU2m4uGAi5QF4qr3dqbbB ys6i1d4x/kuPvfFvZVrsiFVO0EKK5EzunccxHfly4aOpzksJlTCSrKSWoDwltpOK F9UKww53JTPbfNTMmqnOx9sE6DuPz3mNA7DMTMBiPjhl9nouq4mGFma60up2XY/u Ge9qOE7Wrrn95H9JZAR3uQqi5CehnAFd9h2fC2mcyAGkppJ69oOLhIBi/z94hC0= =Fygz -----END PGP SIGNATURE-----

1 0

T Shits for Relay Operators
by Abhiram Chintangal 02 Jun '15

02 Jun '15

Hello, Is the tshirt email system currently working? I have been running a relay since December and I haven't seen one. Btw, I did register my relay using tor weather. Thanks! -- Abhiram Chintangal

2 1

Re: [tor-dev] onionoo: bug in family set detection?
by teor 02 Jun '15

02 Jun '15

> Date: Mon, 01 Jun 2015 19:20:32 +0000 > From: nusenu <nusenu(a)openmailbox.org> > > Hi, > > by comparing different methodologies of "parsing" myfamily data I > stumbled upon differences between onionoo and compass. > > After manual review I assume there is a bug in onionoo (or onionoo has > a different opinion on what families actually are) > > Example: > > According to onionoo, torpidsDEevanzo [1] is part of a family with 38 > members. > > It lists torpidsFRonline [2] as one of its members, that implies that > torpidsFRonline lists torpidsDEevanzo as one of its members as well, > but torpidsDEevanzo does _not_ list torpidsFRonline (according to > onionoo data). > > grep 'fingerprint":"0C77421C890D16B6D201283A2' details.json|grep > 5510FC1736B16D46D3F2DDA5011995C478D42594 > (no result) > > Is this a bug? > > thanks > > [1] > https://atlas.torproject.org/#details/5510FC1736B16D46D3F2DDA5011995C478D42… > > [2] > https://atlas.torproject.org/#details/0C77421C890D16B6D201283A2244F43DF5BC8… No, it's a feature :-) MyFamily requires bidirectional declarations to be effective. This prevents a malicious relay nominating significant portions of the Tor network as its family, in order to direct traffic to another malicious relay. (And/or slowing down the network and attempting to cause a DoS.) In this case: torpids relays have inconsistent MyFamily configurations. OnionOO appears to correctly implement the bidirectional MyFamily logic, and remove inconsistent one-way MyFamily declarations. Compass appears to believe each relay's MyFamily claims, without checking the other relay. This appears to be a fairly harmless bug in Compass, as Compass itself is not used for path selection. teor teor2345 at gmail dot com pgp 0xABFED1AC https://gist.github.com/teor2345/d033b8ce0a99adbc89c5 teor at blah dot im OTR D5BE4EC2 255D7585 F3874930 DB130265 7C9EBBC7

2 1

Re: [tor-dev] The Onion Name System (OnioNS)
by teor 02 Jun '15

02 Jun '15

> Date: Mon, 18 May 2015 15:48:50 -0800 > From: OnioNS Dev <kernelcorn(a)riseup.net> > > ... > > I introduce several data structures, but the most important one is the Pagechain, a distributed structure of linked Pages. Pages contain Records, Records contain .tor -> onion associations. Anyone who is familiar with blockchains will recognize the behavior and application of this structure immediately. However, here the head of the Pagechain is not managed by miners, but rather by a short-lived subset of Tor nodes called a Quorum. They receive Records and merge them into the Pagechain. At the moment I've decided to use 127 Quorum members and rotate them every week. They are randomly selected, but the process is deterministic; I am using the cached-certs + cached-microdesc-consensus documents, which everyone has, to seed a PRNG that then derives the Quorum. What's to stop a sybil attack where the malicious relays try to occupy likely site(s) for the next Quorum? Is the consensus unpredictable enough to thwart this attack? Even during quiet times? (Does Tor have quiet times?) Can we ensure the Quorum servers have to be long-lived and high-capacity (for example, guards) to make it harder to spin up servers and immediately be added to the Quorum? I'm not convinced the Stable flag is hard enough to get. Of course, there's a trade-off where making the set of Quorum Candidates too small makes the Quorum easier to predict, too. By the way, in your ACM paper 5.4.2 you switch from Charlie to Alice, but I think they're meant to be the same Quorum Candidate. > Clients don't need a copy of the Pagechain to use the DNS, but rather they can just rely on their existing trust of the Tor network (including the Quorum and name servers) and verify their signatures on data structures. > Also unlike a blockchain, my Pagechain has a finite length: the oldest Page will eventually drop off, which forces domains to be renewed periodically. I have also introduced mechanisms that 1) allows clients to authenticate the domain name to the hidden service, 2) allow clients to authenticate a denial-of-existence claim from a name server, and 3) prevent name servers from forging .tor -> .onion associations. These vulnerabilities are still generally open on the Internet DNS. I have also tried to minimize networking costs, since Tor circuits are slow. > > To reduce CPU and network requirements, I want Tor routers to have Ed25519 keys. Let this project add additional pressure on that item on the to-do list. > > Recommended readings: > http://onions55e7yam27n.onion -- the official hidden service for this project, but a work in progress. > https://github.com/Jesse-V/Thesis/blob/master/conference/acm-ccs.pdf -- the ACM paper pending peer review > I no longer recommending reading my original thesis, please use the above links instead. > The HS is down, and has been for some days. For those having trouble accessing the paper through the direct link, try clicking on acm-ccs.pdf on: https://github.com/Jesse-V/Thesis/tree/master/conference I got a format error from GitHub when I tried the direct link in Tor Browser 4.5. > … > > I am asking for help with the client-side functionality. I'm currently doing the *.tor interception and lookup resume in connection_edge.c but the software frequently crashes with this approach, (I've learned why) and I'd like to migrate it to Stem for now. I need to intercept .tor domains, pause the lookup (letting the Tor Browser spin), send the hostname over a named pipe or TCP socket, read back a .onion address, then tell Tor to resume the lookup under the .onion address. This way, the HS loads under a .tor domain. I have two questions about the Stem-based approach: 1. Applications which use Tor via SOCKS won't be able to use .tor domains without using Stem. This adds another dependency to apps which want to use .tor addresses, and confuses users by making .tor addresses work with some torified apps, and not others. Is your final goal to have .tor lookups contained within the Tor client? I'm sure someone could help with the C coding if that's the issue, particularly given a working implementation in Python/Stem. 2. I always feel nervous when people say "letting app X spin". I assume Tor Browser just looks like it's doing a DNS lookup or similar? teor teor2345 at gmail dot com pgp 0xABFED1AC https://gist.github.com/teor2345/d033b8ce0a99adbc89c5 teor at blah dot im OTR D5BE4EC2 255D7585 F3874930 DB130265 7C9EBBC7

1 1

Proposal 245: Deprecating and removing the TAP circuit extension protocol
by Nick Mathewson 02 Jun '15

02 Jun '15

Filename: 245-tap-out.txt Title: Deprecating and removing the TAP circuit extension protocol Author: Nick Mathewson Created: 2015-06-02 Status: Draft 0. Introduction This proposal describes a series of steps necessary for deprecating TAP without breaking functionality. TAP is the original protocol for one-way authenticated key negotiation used by Tor. Before Tor version 0.2.4, it was the only supported protocol. Its key length is unpleasantly short, however, and it had some design warts. Moreover, it had no name, until Ian Goldberg wrote a paper about the design warts. Why deprecate and remove it? Because ntor is better in basically every way. It's actually got a proper security proof, the key strength seems to be 20th-century secure, and so on. Meanwhile, TAP is lingering as a zombie, taking up space in descriptors and microdescriptors. 1. TAP is still in (limited) use today for hidden service hops. The original hidden service protocol only describes a way to tell clients and servers about an introduction point's or a rendezvous point's TAP onion key. We can do a bit better (see section 4), but we can't break TAP completely until current clients and hidden services are obsolete. 2. The step-by-step process. Step 1. Adjust the parsing algorithm for descriptors and microdescriptors on servers so that it accepts MDs without a TAP key. See section 3 below. Target: 0.2.7. Step 1b. Optionally, when connecting to a known IP/RP, extend by ntor. (See section 4 below.) Step 2. Wait until proposal 224 is implemented. (Clients and hidden services implementing 224 won't need TAP for anything.) Step 3. Begin throttling TAP answers even more aggressively at relays. Target: prop224 is stable. Step 4. Wait until all versions of Tor without prop224 support are obsolete/deprecated. Step 5. Stop generating TAP keys; stop answering TAP requests; stop advertising TAP keys in descriptors; stop including them in microdescriptors. Target: prop224 has been stable for 12-18 months, and 0.2.7 has been stable for 2-3 years. 3. Accepting descriptors without TAP keys. (Step 1) Our microdescriptor parsing code uses the string "onion-key" at the start of the line to identify the boundary between microdescriptors, so we can't remove it entirely. Instead, we will make the body optional. We will make the following changes to dir-spec: - In router descriptors, make the onion-key field "at most once" instead of "exactly once." - In microdescriptors, make the body of "onion-key" optional. Until Step 4, authorities MUST still reject any descriptor without a TAP key. If we do step 1 before proposal 224 is implemented, we'll need to make sure that we never choose a relay without a TAP key as an introduction point or a rendezvous point. 4. Avoiding TAP earlier for HS usage (Step 1b) We could begin to move more circuits off TAP now by adjusting our behavior for extending circuits to Introduction Points and Rendezvous Points. The new rule would be: If you've been told to extend to an IP/RP, and you know a directory entry for that relay (matching by identity), you extend using the node_t you have instead. This would improve cryptographic security a bit, at the expense of making it possible to probe for whether a given hidden service has an up-to-date consensus or not, and learn whether each client has an up-to-date consensus or not. We need to figure out whether that enables an attack. (For reference, the functions to patch would be rend_client_get_random_intro_impl and find_rp_for_intro.)

1 0