[tor-project] Network team meeting notes, 17 Sep 2018
Nick Mathewson
nickm at torproject.org
Mon Sep 17 17:51:38 UTC 2018
Hi!
Our logs are at
http://meetbot.debian.net/tor-meeting/2018/tor-meeting.2018-09-17-16.58.html
And here are our notes:
= Network team meeting pad! =
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!
(See https://lists.torproject.org/pipermail/tor-project/2017-September/001459.html
for background.)
After each week's meetings, the contents of this pad will be sent to
tor-project @ lists.torproject.org. After that is done, the pad can
be used for the next week.
== Previous notes ==
(Search the list archive for older notes.)
6 Aug: https://lists.torproject.org/pipermail/tor-project/2018-August/001936.html
13 Aug: https://lists.torproject.org/pipermail/tor-project/2018-August/001947.html
20 Aug: https://lists.torproject.org/pipermail/tor-project/2018-August/001954.html
27 Aug: https://lists.torproject.org/pipermail/tor-project/2018-August/001963.html
4 Sep: https://lists.torproject.org/pipermail/tor-project/2018-September/001965.html
10 Sep: https://lists.torproject.org/pipermail/tor-project/2018-September/001972.html
== Stuff to do every week =
* Let's check and update the roadmap. What's done, and what's coming up?
url to roadmap:
https://docs.google.com/spreadsheets/d/1Ufrun1khEo5Cwd6OwngERn829wU3W3eskdrriaYfUBQ/edit#gid=856122210yy
* Check reviewer assignments at
https://docs.google.com/spreadsheets/d/1Ufrun1khEo5Cwd6OwngERn829wU3W3eskdrriaYfUBQ/edit#gid=95993070
* Also, let's check for things we need update on our spreadsheet! Are
there important documents we should link to? Things we should
archive?
* Check rotations at
https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/TeamRotations
* Community guides, it's time to hand off to the next guide!
* Let's look at proposed tickets! [but see discussion]
https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&keywords=~035-roadmap-proposed&max=200&col=id&col=summary&col=milestone&col=status&col=type&col=priority&col=component&order=priority
== Reminders ==
* 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
work for the next releases
-------------------------------
---- 17 September 2018
-------------------------------
== Announcements ==
Tor 0.3.2 will not be supported past 9 October (3 weeks from now):
https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases
== Discussion ==
* Is anything blocking 0.3.5.1-alpha?
teor says:
#27288 makes tor exit if it has a consensus that requires LinkAuth=1:
See: https://lists.torproject.org/pipermail/tor-dev/2018-September/013434.html
We fixed the issue for OpenSSL tors (#27734), but NSS tors are
still affected.
The public Tor network can still generate consensuses that require
LinkAuth=1:
- Consensuses need an absolute majority of configured authorities (5/9).
- Required protocols need a 2/3 supermajority of voting authorities (4/5).
- Currently, 4/5 authorities are on 0.3.3, and still require LinkAuth=1.
- Maybe we should block 0.3.5.1-alpha until another authority upgrades to
0.3.4.8?
- But 4/9 authorities being down is really rare.
[dgoulet: Last week I emailed dir-auth@ list on moving towards a
new stable so we can vote that out.
As of now the consensus does *NOT* have "LinkAuth=1"]
Even then, NSS Tors will exit at launch if they have a cached
consensus which
requires LinkAuth=1 (#27735):
- Maybe we should block 0.3.5.1-alpha until NSS works with an (old) cached
consensus?
- But in a week, cached consensuses that old will be really rare.
Here's what I think we should do now:
- Ask the remaining 4 authorities to upgrade to 0.3.4.8 alpha.
- Try to write a draft proposal for the old cached consensus issue.
- Wait a week after the last consensus that required LinkAuth=1,
then release
0.3.5.1-alpha.
Here's what I think we should do before the release candidate:
- wait for the required supermajority of authorities to upgrade
(6/9). [DONE]
- release at least one 0.3.5 alpha with the old cached consensus fix.
* What sessions do we need for our team day in Mexico?
- See https://trac.torproject.org/projects/tor/wiki/org/meetings/2018MexicoCity/DailyAgenda
- Retrospective & Revisiting our weekly routines (60 min)
- Roadmapping (90-120 min)
- Refactoring plans, next steps, open problems? (60-90 min)
- Rust plans, next steps, open problems? (60-120 min)
- Improving our process -- capacity estimation? (30-60 min?)
* Let's defer non-essential tickets from the 0.3.5.x-final milestone
== Updates ==
Nick:
* Last week:
* Stable releases
* Got OSS-Fuzz build working. (required initialization in
ed25519 code. e5601f14ed4)
* Reviewed and merged lots and lots
* Defered a bunch of feature tickets to 0.3.6 or unspecified.
* Wrote code to make networkstatus mmapped, then fuzzed it,
added its testcases to the corpus, and deferred it to 0.3.6 (#27244)
* Handled some lingering missing-parts of NSS (chutney compat,
byte-counting, closing sockets correctly)
* Tried to get jenkins passing again
* Fixed a test issue where leaks in a tinytest subprocess
wouldn't cause the tests to fail (#27658)
* Fixed some outstanding memory leaks (#27684, #27685)
* Handled some 32-bit apple-os time-computation bugs. (#27139)
* Wrote a patch to fix up our assert() macros so that the
compiler doesn't let you assert(a=b).
* Miscellaneous bugfixes
* This week:
* Revise whitepaper
* Release 0.3.5.1-alpha
* Defer more items from 0.3.5.1-alpha
* Hunt bugs in 0.3.5.1-alpha
* More prep for mexico
* Identify remaining roadmapped items in advance of mexico meeting
* Write a proposal about better handling required-but-absent
protover stuff, to prevent another problem like #27734
haxxpop (offline):
last week:
- Nothing IIRC.
this week:
- Maybe we should discuss about the cli for client auth keygen
tool. Should it be `tor-keygen`? `tor --keygen`?
IMO, I think we should implement the client keygen tool in C
because currently Rust is not required by default.
I'm afraid that the tool will not be included in the binary if it
doesn't --enable-rust.
[dgoulet: Agree and we have all the utils/libs in the Tor code
base for this tor-keygen tool which could go in src/tools]
teor (offline):
last week:
- Attended the relay operators' IRC meeting
- Reviewed #27247 and #27629 for Sponsor 8 RAM usage
- Worked out the available compiler versions in our CI for #27468
- Got a response from Travis CI about startup hangs and network
failures (#27366),
it was confusing, so I asked them to clarify
- Diagnosed "required protocol" errors (#27288)
- Reviewed a few protover tickets
- Took some leave
this week:
- Do travel preparation (so many pieces of paper!)
- Catch up on meeting planning
- More Reviews! (Try to help with the 0.3.5 feature freeze backlog)
- Install our Travis macOS dependencies more reliably, and
faster (#27738)
- Make our CI run the latest GCC and clang (#27468)
- Add NSS to our CI (#27751)
- Create an updated list of fallback directory mirrors (#24786)
ahf
Last week:
Sponsor 8:
- Hacked on #25502 tests and ran into some troubles with my
timer-based event loop handler for Windows, Unix parts seems to
work fine now, but I'd like to make sure this works the
same way on both
Windows and Unix since we change the way transports.c is
reading data from the
PT worker processes.
Misc:
- Reviewed: #27289, #27451, #27664, #27685, #27684
- Initial (but unfinished) reviews of #27194, #27315, #27709
- Wrote a test implementation of a service that uses the feature in
#4700 for testing:
https://github.com/ahf/charybdis/commit/15781f5a22e82dd8a963c0f032e0ae1c8b725647
- Fixed some issues with our patches for #4700.
- Participated in a podcast about Tor.
- Went over some press material for Steph about CF.
- Went over Roger's big draft proposal.
This week:
Sponsor 8:
- Get #25502 issues fixed and uploaded for review.
Misc:
- Work on bugfixes/reviews for 0.3.5 branch(?)
catalyst:
last week (2018-W37):
- #27169 (defer dirinfo bootstrap progress) - made pull
request and merged
- more work on bootstrap reporting items
- talked with arma some about PT + proxy stuff
- looks like work for #27100 (add PT connect bootstrap status)
and #27167 (track first OR connection) will overlap significantly
this week (2018-W38):
- (bug rotation?)
- reviews
- travel prep
- 0.3.5 bugfixes as needed
- more work on #27100 and #27167
asn:
Last week:
- Finalized #4700 with the help of ahf. It's now in merge_ready.
- Fixed some coverity issues with v3 client auth
- Reviewed #27678 and #27464.
- Received email from a few v3 operators telling me they are experiencing
reliabiliy issues. Spent lots of time debugging the logs. Some of them look
like #25882, others look like potentially fixed bugs in master, while others
might be actual issues we don't know of yet.
This week:
- Work on onion proposal.
- Work on SponsorV report.
- Possibly debug more v3 logs...
- Reviews.
dgoulet:
Last week:
* Worked and reviewed on many 035 tickets, rushing the pre-freeze.
* Ticket triage on 035.
This week:
* Bug hunting and squashing for 0.3.5 as we enter the post-freeze and
036 pre-merge-window.
Mike:
Last week:
* Worked with dgoulet on fix for bw stats leak (#23512)
* Fixed up #27678
* Reviewed tickets
* Started work on travel safety guide
This week:
* Finish travel safety guide
* Update vanguards for 0.3.5.1
More information about the tor-project
mailing list