[tor-project] Network team meeting notes, 26 March 2018
nickm at torproject.org
Mon Mar 26 17:58:44 UTC 2018
Our weekly meeting logs are available from:
Below is a record of the contents of our pad.
= Network team meeting pad, 26 March 2018 =
ROS: My name is Guildenstern, and this is Rosencrantz.
(GUIL confers briefly with him.)
(Without embarrassment.) I'm sorry - his name's Guildenstern,
and I'm Rosencrantz.
-- Tom Stoppard, _Rosencrantz And Guildenstern Are Dead_
Welcome to our meeting! Mondays at 1700 UTC on #tor-meeting on OFTC.
(This channel is logged while meetings are in progress.)
Want to participate? Awesome! Here's what to do:
1. If you have updates, enter them below, under your name.
2. If you see anything you want to talk about in your updates, put
them in boldface!
3. Show up to the IRC meeting and say hi!
Note the meeting location: #tor-meeting on OFTC!
== Stuff to do every week ==
* Let's check and update the roadmap. What's done, and what's coming up?
* Check reviewer assignments at
* Check rotations at
== Announcements ==
* 0.3.3.x triage has happened. How was it?
* Starting 0.3.4.x triage this week. I could use help.-N
* Remember to "/me status: foo" at least once daily.
* Remember that our current code reviews should be done by end-of-week.
* Make sure you are in touch with everybody with whom you are doing
* Important dates:
* Apr 15, 2018 -- 0.3.3.x stable is supposed to be ready!
* May 15, 2018 -- 0.3.4.x feature freeze!
* Dgoulet now has commit permissions cool! :confetti_emoji:
* dgoulet will be AFK on Friday/Monday, national holidays (Easter...).
* ahf will be semi-AFK on Thursday/Friday/Monday (will be around for
the meeting for sure) for easter.
* isis will be semi-AFK on Friday/Monday
* 0.3.3.x triage status:
* 033-included-2018 are the ones that are still included after the
first 2 rounds:
* There are also some new tickets still in the milestone that
survived the first round of triage, but which we aren't currently
planning to tackle in 0.3.3.x-final. They are tagged with
== Discussion topics ==
* Kat reminds us that we should nominate volunteers to receive Tor swag.
- Let's maybe do this on another pad, so we aren't discussing
volunteers in public :/
* We need to finish all the 033-must tickets. How are we going to do this?
= Updates =
== Nick ==
* Lots of post-rome tasks
* Did 0.3.3.x triage
* Worked on lots of little bugs
* Helped interview 4/5 of ED candidates. Last 1st-round interview
* Went to Libreplanet on Saturday, gave a quick overview of
network team stuff as part of State Of The Onion, talked to bunches of
people. (Thanks to isabela and stephw and flexlibris and n8fr8!)
* Cleared out my review queue
* Quick refactoring branch to help people experiment with other
relay crypto designs.
* PETS reviews, week 1.
* Tons of meetings
* Putting out 0.3.3.4-alpha.
* 033-must tickets
* Talking with dgoulet about CPU-usage-when-idle (done)
* Talking with ahf about idle API
* Talking with dgoulet, isis, catalyst about wide creates
* catch up on all the follow ups from Rome / sent summary to the ml
* prep for this meeting
* follow up with FB proposal and censorship team
* Post-Rome tasks: notes to the Wiki, emails.
* Triage 0.3.3 tickets.
* Addressed 033-must tickets. Some important ones: #25226 and #24767.
* Reviewed and opened some tickets. See Timeline for this:
* Organized our roadmap modularization work with ahf: #25498.
* Organized our roadmap Reduce CPU usage work with nickm: #25500
* Feedback on mike's tor-dev@ email thread: [tor-dev] Setting
* Helped breaking down Sponsor Bucket network team tasks with asn.
* Ticket work and review, most of it will be 033-must.
* Modularization work with ahf.
* Follow up on larger tickets from last week (#25226, #24767).
* Organize the work for wide CREATE cells with isis/nickm/catalyst.
* Mailinglist posts about two guards & QUIC
* Meetings about other funding proposals, personel
* (re)read a handful of research papers, filed tickets
* Some ticket review
* Xfer two guards thread -> proposal. Does anyone have any other
* TL;DR is that one guard is not any better than two, and
arguably worse, because our path restrictions allow an adversary to
force the use of a second guard at any time for specific activity. I
am assuming we want to keep our path restrictions. Is this correct?
* Relatedly, in order to prevent the adversary from being able
to force the use of a *third* guard, we also need to ensure that the
two guards we pick are not covered by the same path restrictions (ie
must be chosen from different /16 and family).
* Trim down prop247 to cover what we are doing with the vanguard
script (which will also match our planned final implementation).
* Review other tickets. (Are we deliberately assigning a new
person to review a ticket each time it goes back to needs_revision?
This seems expensive, but does mean more eyeballs)
[dgoulet]: No, if it goes back in needs_revision, the original
Reviewer stays as is and in charge of reviewing again when it goes
back in needs_review.
last three weeks?:
- finding bugs #25517
- refactoring protover and giving it a more memory efficient
voting algorithm #24031
- reviewed linker/build modifications for testing rust code
which calls C code #25386
- adding more tests for previously untested things #25605 #25425
- giant email to Ian about various handshake design
considerations and (mis)usage(s) of hash functions
- revising prop#269 based on said discussion/reading
- documented our current policy for Rust dependencies and set
up a script to help vendor them #25310
- reviewed a unittest that nick and juga wrote #25515
- ticket/hackerone triaging while people were in rome/afk
- reviewed volunteer win32 patches from mailing list #25479
- reviewed/shepherded new rust module for calling tor's logger #23881
- fixed a flaky HSv3 unittest #25450
- reviewed other changes to Rust coding standards #25368
- wrapped our sha2 implementation so that we can call it from
- reviewed teor's crypto_rand_double_sign() in rust #25381
- reviewed/shepherded next stage of the crypto.h refactor #24658
- bought a mac and set up stuff like fish and rustup and git
for testing thingsxk
- wrap our RNG? #24660
- finish the sha2 thing since it's also blocking the handshake
- more prop#269 thinking/writing
- Discussed basic modularization with David (#25498):
- Started looking into building a Tor with some of the
dirauth code ripped out.
- Reviewed: #25398, #25399, and #25512.
- Post-Rome tasks: emails, backlog, look at tickets.
- Wrote the onion_status.pl script for status reporting.
- Got the subprocess API to work with the event loop on Unix. Now
trying to get it work on Windows.
- Participated in the weekly PT meeting.
- Continue with trying to build a Tor with some dirauth code left out.
- Discuss Idle interface with Nick.
- Need to review #25425 and #23846.
- Bug triage rotation role.
Last week: Nothing
Next week: I will start implementing the client authorization for v3
onion service (as I told dgoulet and asn in Rome)
* collecting questions for the virtual "bandwidth authority team"
* started merging PRs without review
* solved some bugs and PRs
* cleaned up not used code
* status: still buggy and not getting nice numbers
* sbws: still not sure whether to dedicate time on it or continue
* got access to ln5 box for testing, though still not running
bwscanner to do not generate traffic
* received negative answer from previous funding attempt (aka F1)
* submitted application for new funding attempt (aka F2)
* writing application for new funding attempt (aka F3)
* Look sbws more in detail, document?
* solve current blocking bugs:
(https://github.com/TheTorProject/bwscanner) u#111, #94, others
* document better what is doing
* try run it with shadow?
* funding: submit application for F3 (deadline sat)
* time for working on #25515?
* i would like to join this meetings to continue sharing updates and
discuss bwscanner questions when i have them. - is this cool w/ the
team?, and where/when/who i could send questions that would talk
longer to discuss?
sbws (simple bw scanner) is my attempt as a bw scanner
been running some scans primarily for timing information
got set up on ln5's donated box
come up with and implement an intelligent scan order for relays
last week (2018-W12):
- day off to recover from travel
- dealing with contractors at home
- wrote up notes from Rome
- sent near-term recommendations to team from CI session
(thanks everyone who started signing up for and enabling their
- reviewed tickets #24740, #24659
- installed Homebrew and Rust on my Mac to help review stuff
that needs those
- helped people with Travis and GitHub setup
- CoC and SoV feedback
this week (2018-W13):
- talking to people about CC stuff
- more ticket review
- help with 033 and 034 tickets as needed
- Coverity rotation duty
Last three weeks:
- Attended Rome meeting!
- Gave Tor talk in ENS Paris. Gave talk in Tor meetup hackfest in Athens.
- Worked on network team roadmapping and worked on spreadsheets for
roadmap and metadoc.
- Worked on SponsorBucket roadmapping and briefed team about
upcoming Snowflake position.
Need help from Roger/Isa on figuring out next steps based on my
last email to the thread.
- Handed over the first round of weekly reviews with David.
- Reviewed #17799, #24767, #24989, #6236.
- Setup github account, github repo and travis and started
transitioning there: https://github.com/asn-d6/tor
- Discussed switching to 2 entry guards with Mike on [tor-dev]. I can be
persuaded that it's a good idea but we need a proposal.
- Worked on #14389 (UX of client auth for onions) with Arthur and
David. Need more little-t-tor work and analysis.
- Opened ticket about rev counter issue found by Roger and made an
implementation plan: #25552
- More reviews.
- Start working on my vanguard simulation again. My goal is to have something
initial on github by the end of the week, so that Mike can start using it
in a week or two.
- Need to do more work on #20212 and #25609 so that we can move forward
there. Some mikeperry wisdom might be needed there.
More information about the tor-project