[tor-project] Network team meeting notes, 11 Sep 2017
nickm at torproject.org
Tue Sep 12 14:53:08 UTC 2017
Hi! Our meeting log is here:
and below is the contents of our meeting update pad:
Network team meeting, 11 September 2017 (1700 UTC)
Announcements and requests:
- Can we have trac back? (It was down Sunday/Monday UTC)
[It's back up now. Folks are working to try to make it more
reliable. see thread.]
- metrics team request?
- privcount/prop280 meeting tomrrow (er, 2017-09-13 00:00Z)
- Last week or two:
* Released (experimental) PrivCount 1.1.0: https://github.com/privcount
* Kept at it with the floating-point stuff. Now we have dev guidelines
(thanks catalyst!), and a bunch of specific issues:
* Work out how many onion addresses an HSDir should expect to see
(and intro and rend weights as well):
[asn: interesting stuff! definitely connected to #23126 as
well. we should think
of what to do for v3. perhaps montreal session] +1,
including thinking of useful stats with privcount.
[dgoulet: The overlap period concept has changed also re #23367]
* Revised the patch that fixes bridge clients on tor master
* Reviewed a bunch of nickm's spec patches
- This week:
* Start developing PrivCount 1.2.0 (HSDir client usage)
* Try to knock over a few of the floating-point bugs
* Merge HSv3 support into chutney (thanks asn, dgoulet!), and add a
v3 test network to tor master's make test-network-all
* Last week
- Release 0.3.1.6-rc
- Review and merge lots of patches, cleaning up as needed.
- Run through trivial open 0.3.2.x tickets, try to solve the fast ones
- Some triage
- Get callgraph generation working again. (See
- Looked over some Rust stuff with Komlo.
* This week
- 0.3.2.x freeze: merging, coding, reviewing, triaging. Trying not to
be too ambitious.
- Triage plan: if there is anything in 0.3.1 or 0.3.2 you will not do,
and it is assigned to you, unassign it, please. If it is essential,
please talk to others about making it happen.
- Triage plan: i'm likely to remove (nearly?) all stuff in 0.3.1 or
0.3.2 that is unowned. If you will do it, pleas assign it to
- Try to do a little 0.3.3.x planning-planning
- Think about 0.3.1.x: why so late?
- Start on a couple of proposals for 0.3.4.x and forward.
- Privcount meeting tuesday night
* Last week
- spent most of the week working with Linda and Steph (f2f here in
NYC) on the dev and community portals part of the website redesign
- the only deliverable network team had with 4 is now officially 'close'
* This week
- organize sponsor8 work
- start preparing for roadmap in Montreal with the team
- how are we with dependencies for TB team on Tor Launcher work?
moat api work and the progress bar requests from mcs
* Last week:
- Most of my time was hunting down #23387 and working on a fix and unit
- Worked on KIST branch addressing pastly's review of #12541. It is now on
oniongit awaiting review and hopefully make it in 032.
- Reviewed some tickets in review-group-23.
* This week:
- Hunting down prop224 bugs and writing tests.
- Will work on more general Tor for 031 and 032 upcoming releases.
* Last two weeks (2017-W35, 2017-W36):
- Went down a floating point rabbit hole. Reviewed some stuff
(#23061); wrote some guidelines (#23368).
- Reviewed some patches, including the large buf API refactor (#23149)
- Looking into bootstrap reporting led to a twisty maze of periodic events
* This week (2017-W37):
- Work on control protocol extensions for bootstrap reporting. Extend
OR_CONN events to report per-connection PT progress? Or create a new
class of events?
- Maybe keep an eye on privcount stuff as relates to floating point.
- Extract bootstrap reporting from deep in the call stack and make it
the responsibility of higher level code.
- Better error message reporting to help Tor Launcher
- Back from vacation tasks: read up on email threads, time sheets,
meeting logs, trac, tor.git commits, signed up for team rotation
- Got travel approval and booked trip for the Montreal meeting via
Jon+Shari. Arriving the 6th of October.
- Send traveling information to OTF for the OTF meeting in Valencia.
- Continuing to keep the NG HS IRC running: found one bug
legacy and NG HS on the same tor instance this week.
- Working on a Tor instance where I can send events from the
Android Java interface to Tor via orbot.
- Unbricked a device that I had borked after a failed attempt at a
self-build Android image with some debugging enabled.
- Going over Nick's event list from August about timed event
sources and wrapped my head around that.
- Help out with 0.3.1 release if need be?
- Continuing with the sponsor 8 work:
- Try to flesh out a plan with some deadlines for the
tasks/results. Maybe together with Isa?
- How's it going with the Android OS dev hire we were looking for?
- Spend some time wrapping my head around Rust before the montreal
- Meeting in Seattle
- Page back in everything since vacation. Here's the dump from my notes:
- #23077, #23097; lower prio: #22934, #23100, #23101
- Design + document planned prop247 prototype performance experiments
- Worked on fixing more prop224 bugs that have crawled up (#23331, #23056,
#23327, #23355, #23361). We've merged most of the fixes upstream by
now. The current tor.git master offers a pretty good prop224 experience. No
blockers are currently known.
- Worked on debugging and fixing a serious reachability issue of prop224
which could cause reachability issues under natural causes: #23387. We
identified it while writing some advanced unittests for prop224 (see
#23310). The fix required both a spec patch and a code patch which has
already been merged mainline.
- Got more people to try prop224 and join our experimental ircd server. The
testing has so far been very useful since with more people we can find more
rare bugs. We've already caught #23455 this way and have learned more info
- Get #23457 and #23455 fixed.
- Test and fix more bugs.
- Get into review-group-23.
- more work on #15967
- trac kept going down and losing stuff i typed :'(
- finish #15967
- start #22871 (moat backend)
Finished a 2 week vanilla/kist experiement using the KIST dgoulet
reviewed. I have graphs.
Coming up: review dgoulet's KIST work again #12541
More information about the tor-project