[tor-project] Network team meeting notes, 5 Feb 2018

Nick Mathewson nickm at torproject.org
Tue Feb 6 02:03:45 UTC 2018


You can see logs from today's network team meeting at


Our notes follow:


Network team meeting pad, 29 January 2018

             Not all bits have equal value.

                  --- Carl Sagan, Cosmos (1980)

Welcome to our meeting! Mondays at 1800 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!
 (See https://lists.torproject.org/pipermail/tor-project/2017-September/001459.html
for background.)

Meeting notes from previous weeks:
    * https://lists.torproject.org/pipermail/tor-project/2018-January/001639.html
    * https://lists.torproject.org/pipermail/tor-project/2018-January/001644.html
    * https://lists.torproject.org/pipermail/tor-project/2018-January/001647.html

Old Announcements:
    * Let's have some proposal discussions. Isis kicked off the process here:
    * On the roadmap spreadsheet: Please take February tasks.  (If
somebody else has already taken something you want, please talk to
them and/or add yourself too.)
    * 0.3.3.x is in feature-freeze.  No new features (except for
#24902, which has permission.).
    * No 0.3.4.x patches will be merged till the window opens on Feb 15.
    * Please, work on bugfixes!  It would be great to have this
release release on time. (Planned date is 15 April)
    * There are tons of tiny 0.3.3.x tickets.  Maybe if you can do one
of them in <X minutes for some small X, you should just do it?

    * Support for 0.3.0.x ended on 1 Feb. (That leaves 0.2.5, 0.2.9,
0.3.1, 0.3.2, and 0.3.3).
      0.2.5 is EOL in May, and 0.3.1 is EOL in July.
    * There are 7 tickets left in needs_review in review group 31
      Please help all your colleagues' tickets get reviewed.
    * May hackfest has been approved by Shari.
      dgoulet, teor, and isabela are organising it off-list.
      See the network-team list for updates.
    * prop#239 discussion meeting on Thursday 8 February at
21:00-22:00 UTC in #tor-meeting

    * The March pre-meeting hackfest day is Saturday 10 March 2018
      We want a single-day pre-planned highly-focused hackfest
      Let's limit the day to 6 hours and 6 sessions?
      Here is a pad for topics and coordinators:
      teor: added IPv6, and komlo's Rust
    * The all teams meeting day is Sunday 11 March 2018
      Current sessions are here: https://ethercalc.org/zil3cf5nm9ge
      teor: added PrivCount in Tor for network team and metrics

    * Last week:
        - Implemented and tested Experimental PrivCount onion service stats
        - Fixed some more experimental PrivCount bugs
        - Did not get time to review or revise much Tor code
        - More reading on floating point, most operations we want are
precisely defined by IEEE 754
    * This week:
        - More Experimental PrivCount features
        - Try to merge the DDoS fixes into Experimental PrivCount
        - Eventually make a trac user page that lists the tickets I'm
working on in priority order
        - Maybe I will get time to review or revise tickets this next week

    Mike, if you're out there, bug 22212 now includes a diagnosis and
a branch that fixes it.

    Last week:
      - Worked on making the vanguards repo more usable (wrote a
README, added some control port options)
      - Contacted Matt Wright et al about adaptive padding work
      - Reviewed the updates to the Tor Browser design doc
    This week:
      - Juggling vanguards, adaptive padding, other things
      - Will comment on #22212

    * Last week:
        - Administriva, Rome planning, etc
        - Worked on 24902 merge, many reviews, many bugfixes
        - And did I mention the bugfixes?
        - Deprecated 0.3.0
        - Tracked down some false positives from clusterfuzz
        - Tried to work somewhat on 0.3.3 ticket volume.
        - Release engineering.
        - Met with mozilla folks to braindump about Tor.  Verdict: our
specs were more complete than expected; they just needed a better
introductory explanation.  We should get to this some day now.
    * This week:
        - I'll be sitting in on a class on proofs & coq (twice a week,
all spring.)
        - So much release engineering. I am not getting much else done.
        - If I can, work on any of the stuff that's been stalled as I
did release engineering (bugfixes, sponsored work, etc.)

  * Last week:
    - Some 033 ticket triage.
    - Went over tickets in the review-group-31 and some tickets as well I was
      the owner.
    - We got the DoS mitigation merged in #24902.
    - Implemented a geoip client cache OOM handler (#25122) which also was
    - Another important fix was #24700 got merged. The scheduler bug #25125 got
      opened after that found by the extra BUG() we added to #24700.
    - Some bugs appeared after all we've merged: #25128, #25148.
  * This week:
    - Continue the 033 review and bug squashing especially with DoS mitigation.
    - I hope to be able to spend time with #24554 scheduler improvements.
    - Performance and memory analysis. We still have relays at insane level of
      RAM and hugging CPU with 032 so I want to spend a bit of time looking at
      this again.

    last week:
        * began scheduling proposal discussion meetings! please vote
for times if you're interested in them!
        * worked on #24986/prop#249 large create cells

        [dgoulet]: If you have a branch for some initial code, would
be good for others to skim it quickly to avoid any big design

                   misunderstanding which would make the whole review
process much more tedious (if you want early feedback that is).

                   This is a _big_ job :) okay will do! thanks! i have
a bunch of places marked with stuff like "XXX am i supposed to do Y
here?" [nickm: We'd love to look at those]

        * some refactoring work in the process to cleanup common logic
related to parsing cells
        * relearning how to write coccinelle scripts (their docs are…
very ungood…)
        * should create2v_cell_body_t be contained within a var_cell_t or?
        * conversations with the CC
        * started trying to shift the Google AppEngine/Cloud/Compute
account for moat to TPI ownership so that we can actually fund it and
use it
        * attempted to diagnose a moat issue with it not finding any
bridges to give #24432
        * contacted a person at Intel about a type 1 spectre static
analysis tool
    this week:
        * more #24986
        * prop#239 meeting on thursday
        * probably start writing PQ KEX proposal #24990
        * try to reschedule meeting with Ian about ntor changes since
that meeting never happened

    Last week
        Sponsor 8:
            - Wrote a Coccinelle "patch" [isis: lol, i feel your pain]
together with Hello71 for adding
              type-information to calls to our allocators to track down
              type-information as part of allocations: example: convert
              tor_malloc(sizeof(T)) to tor_malloc(sizeof(T), STRINGIFY(T)).
            - Analyzed padding "holes" in our C struct's after a chat with Tim.
              Looks like we could save some bytes here and there for some of
              the objects, but nothing spectacular. I think I wont spend more
              time here and create a master ticket with the results and flag it
              as easy. The task is most of the time to "reorder"
members. [nickm: I'm in favor of this, but only for objects that we
allocate hundreds of. eg, no reason to do this for or_option_t.][ ahf:
+1 ]
            - Debugged a platform "issue" on Android where my 8.1 device
              switches between 4g and WiFi. Led me to also read about Google's
            - Read the onion NS proposal.
            - Submitted a patch for #25120.
            - Sent an email to the network team list about GSoC - does people
              have anything to add here?
            - Went to FOSDEM.
            - Booked Rome traveling.
            - Did end of month reporting.
    - This week:
        Sponsor 8:
            - Memory optimization
            - Get back to the baselining of event loop usage that I didn't get
              around to last week.
            - Bug triaging & and update roadmap.
            - Find some missing data (is on people.tpo which is down
right now) for Isa for s8 reporting.

    Last week (2018-W05):
        - troubleshooting Travis Rust failure (#25127)
        - CoC and Statment of Values feedback
        - opened #25120 based on user feedback. looked up some stuff
about safer ways to deal with urandom (thanks ahf!). are we
comfortable with downgrading that warning without having a "safe"
urandom fallback first?
        - still sick (but recovering)
    This week (2018-W06):
        - think about how we want to handle long-standing CI failures
(and maybe write up stuff about it)
        - help with #25127
        - look into the urandom safety aspect of #25120, or open a
separate new ticket about it

    sponosr8 Q4 (was sick last week and couldn't really work on it) -
catalyst do you have tickets i should add to the report related to
s8-errors keyword [catalyst: i can double-check. it's tickets Sep-Dec
2017 for the Q4 report right? yes]
    sponsor4 extension is on way - work agreement suggested by nick
sent out to isis
    roadmap in rome - working on a new process for the team that
involves homework/prepwork to be done in Feb more in the email list
    review user feedack from testing in india

  Last week:
  - Updated onion auto-redirect proposal:
  - Reviewed #24849, #24972, #24976, #23814.
  - Started writing a rough skeleton implementation of prop#279.
    Far from the real thing but might help some volunteers get started
with the project.
  - Did some thinking on #24456 (ripping out guardfraction).
    Seems like a bigger operation than originally thought.
  This week:
  - More reviews and helping the team.
  - Do some thinking on roadmapping for Rome meeting.
  - Continue writing prop#279 skeleton implementation.
  - Get in touch with Mike about prop247. Whatchu cooking mike?

    - i have a bwscanner instance chugging along (despite some
unhandled errors which seem related to lost tcp connections);
      too early to guess how much work it'd be to turn the resulting
data to use in the real network though;
      seems i'm not the only one who cares about this though -- both
juga and tjr seem to be working on this

More information about the tor-project mailing list