14:33:28 <karsten> #startmeeting metrics team
14:33:28 <MeetBot> Meeting started Thu Apr  5 14:33:28 2018 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:33:28 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:33:45 <karsten> https://storm.torproject.org/shared/5h1Goax5eNusxjXJ_Ty5Wl7hFR1uqCReUiN8xdlBG8T <- agenda pad
14:35:28 <irl> no more topics to add from me
14:35:30 <iwakeh> hi there!
14:35:33 <irl> hi!
14:35:34 <karsten> hi iwakeh!
14:35:35 <iwakeh> me neither.
14:35:40 <karsten> okay, great!
14:35:45 <karsten> let's start then.
14:35:51 <karsten> * March report (karsten)
14:35:57 <karsten> I put the draft on the pad.
14:36:05 <karsten> want to take a look and see what's missing/unclear?
14:36:17 <iwakeh> we should add tomorrow's release
14:36:26 <karsten> not in march anymore.
14:36:34 <karsten> it will go into the next report.
14:36:42 <iwakeh> true.
14:36:48 * iwakeh ahead of time :-)
14:36:56 <iwakeh> err behind.
14:38:08 <irl> i added the ticket for the atom feed, and changed the other ticket to use bugs.tpo instead of trac.tpo
14:38:16 <irl> which was a convention used by tor weekly news
14:38:25 <karsten> oh, okay.
14:38:28 <karsten> sure, sounds good.
14:39:02 <irl> i think those are all the changes from me
14:39:48 <karsten> okay. iwakeh, anything you think needs changing?
14:40:51 <karsten> still here?
14:41:08 <iwakeh> all fine.
14:41:10 <karsten> ok!
14:41:27 <karsten> should I wait with sending out this report, or can it go as is?
14:41:33 <karsten> after this meeting, that is.
14:41:38 <irl> i don't think there is need to wait
14:42:09 <iwakeh> That'll be fine.
14:42:14 <karsten> alright. will send it out then.
14:42:19 <karsten> * Roadmap update April 1 (karsten)
14:42:26 <karsten> see pad.
14:42:37 <karsten> I started going through the list of open goals.
14:42:42 <karsten> and added notes.
14:43:19 <karsten> please add yours. focus is what percentage done we can write by april 1.
14:44:33 <karsten> another aspect is which goals we'd drop first if we realize we're running out of time.
14:46:41 <iwakeh> percentages seem ok to me
14:47:03 <karsten> ok.
14:48:20 <irl> ok, i've commented on all the goals that i feel qualified to comment on
14:48:30 <karsten> great!
14:48:43 <karsten> so, I'll update the wiki after the meeting.
14:48:57 <karsten> moving on?
14:49:04 <irl> ok
14:49:10 <iwakeh> ok
14:49:13 <karsten> * Adding irl to groups (karsten)
14:49:21 <karsten> this is about running services.
14:49:35 <irl> so ldap groups?
14:49:37 <karsten> given that we meet next week, should I ask you to be added to groups now?
14:49:52 <karsten> yes. collector, metrics, onionoo, exonerator.
14:49:57 <irl> ok, yes, this sounds sensible
14:50:20 <karsten> for now it wouldn't mean taking over a service, but it would enable you to look.
14:50:24 <karsten> iwakeh: does that make sense?
14:50:41 <iwakeh> yes
14:50:50 <karsten> okay.
14:51:23 <karsten> speaking of,
14:51:24 <karsten> * Aberdeen meeting preparation (karsten)
14:51:32 <karsten> what else should we prepare?
14:51:50 <irl> i've requested a meeting room for the two afternoons
14:52:07 <irl> do you have eduroam credentials or should i also get a wifi account?
14:52:22 <karsten> I don't have such a thing.
14:52:30 <karsten> so, yes, a wifi account would be good.
14:52:32 <irl> ok, i'll get you some temporary eduroam credentials too then
14:52:37 <karsten> great!
14:52:44 <irl> the room will have power sockets and chairs
14:52:49 <irl> is there anything else we need?
14:52:50 <karsten> can you send me the address and time?
14:52:58 <karsten> heh, a converter.
14:53:01 <karsten> (I have one.)
14:53:02 <irl> we have those
14:53:06 <karsten> oh, great.
14:53:16 <karsten> no, I think that's good then.
14:53:28 <irl> ok, i can send you the address/time by email
14:53:34 <karsten> yep.
14:53:37 <irl> but it will be one of the university campuses
14:53:53 <irl> (it may be one day on each)
14:53:59 <karsten> sure.
14:54:01 <irl> oh
14:54:04 <irl> do you need parking?
14:54:07 <karsten> yes.
14:54:13 <irl> ok, i will arrange visitor parking too then
14:54:23 <karsten> perfect!
14:54:46 <irl> updated the pad
14:54:56 <karsten> great! moving on?
14:55:07 <irl> ok
14:55:11 <karsten> * Onionoo 5.2 release (karsten)
14:55:28 <karsten> iwakeh: I updated #25700.
14:55:37 <karsten> which is, I think, the last remaining ticket.
14:55:54 <karsten> once that is in I can prepare the release.
14:55:55 <iwakeh> ok, I'm reviewing 25712
14:55:58 <iwakeh> too
14:56:17 <karsten> okay.
14:56:44 <karsten> so, we're on schedule for releasing tomorrow?
14:56:50 <karsten> what time did we say again?
14:56:56 <karsten> for deploying?
14:56:58 <iwakeh> yep, seems like it :-)
14:57:08 <iwakeh> 12UTC?
14:57:22 <karsten> or maybe 11:50 UTC?
14:57:35 <iwakeh> should work.
14:57:41 <karsten> just because that gives us a bit more time until :15.
14:57:43 <karsten> okay.
14:58:21 <karsten> great!
14:58:28 <karsten> last topic on the list:
14:58:32 <karsten> * An "Is my relay running correctly?" service (irl)
14:58:49 <irl> based on the number of times this question is asked on #tor, i think we should automate this
14:59:05 <karsten> how is that different from relay search?
14:59:13 <irl> usually what happens is arma is looking up in his dirauth whether or not it's seen a descriptor and is reachable
14:59:17 <irl> arma does not scale
14:59:27 <irl> it's different because it's instant feedback
14:59:44 <irl> an initial version maybe just does a tls handshake and is happy with it
15:00:00 <irl> a better version does begindir and fetches a descriptor, tests ipv6 and suggests filling in a contact line if not present
15:00:18 <irl> this is going to be a new ideas ticket, so i just wanted to raise it here first before filing it in case there were more ideas
15:00:40 <karsten> so, we offer a service that makes connections to relays.
15:00:44 <irl> yes
15:00:55 <karsten> does it first check whether the provided IP address is a relay?
15:00:59 <irl> no
15:01:06 <irl> because it's meant to be used to diagnose faults
15:01:12 <karsten> can that be exploited?
15:01:28 <irl> https://www.ssllabs.com/ssltest/analyze.html?d=at-tor-1.erg.abdn.ac.uk&s=
15:01:35 <irl> there are already services that will make connections
15:01:44 <irl> this is what gave me the idea
15:01:51 <irl> we would need to put rate limits in though
15:02:21 <irl> if it doesn't to tls we would abort, if it's not a tor relay when we connect we would abort
15:02:25 <karsten> or can we answer this by using more recent tor data?
15:02:57 <iwakeh> wait for another consensus?
15:02:59 <karsten> like, by asking the dirauths if they know about a server.
15:03:14 <irl> we could do that, but it won't diagnose faults
15:03:16 <karsten> that's less instant, but yes.
15:03:31 <irl> e.g. ipv6 advertised in descriptor but it's firewalled
15:03:36 <iwakeh> well, people could be nformed abot that.
15:03:39 <irl> so it never goes into the consensus
15:04:07 <iwakeh> no, but the descriptor might be there.
15:04:27 <irl> we could check that too, but i think that would be a complimentary check as opposed to a replacement
15:04:40 <irl> almost every day i see these questions in #tor
15:04:45 <karsten> okay, this sounds like a fine ideas ticket.
15:04:54 <irl> (i get highlighted on mentions of atlas, so i get to see them all)
15:04:57 <karsten> but I'm concerned about making requests triggered by users.
15:04:59 <iwakeh> I think people can be warded off, for a few hours.
15:05:15 <iwakeh> if that is documented in tor and communicated.
15:05:48 <karsten> maybe we can answer some of these questions by explaining better (on relay search) what we checked by looking at descriptors and what we found out.
15:06:06 <karsten> also, maybe votes tell us more here.
15:06:11 <iwakeh> that's also a good option,
15:06:15 <irl> votes would be very useful
15:06:40 <iwakeh> most likely only the minority of relay operators-to-be ask on irc.
15:07:23 <irl> we should probably improve the "no such relay" page on relay search to point to the community portal documentation once it exists
15:07:27 <irl> on how to debug your relay
15:07:37 <karsten> yes, sounds good.
15:08:15 <irl> ok, i'll turn all this into a ticket then
15:08:21 <karsten> thank you!
15:08:28 <karsten> it's good to turn this feedback into ideas.
15:08:38 <karsten> feedback from relay operators, that is.
15:08:54 <karsten> alright. out of topics!
15:09:10 <irl> oh
15:09:15 <karsten> oh!
15:09:25 <irl> * Metrics/Analysis Tickets (irl)
15:09:36 <karsten> right.
15:09:56 <karsten> see also https://trac.torproject.org/projects/tor/ticket/6473#comment:7
15:10:04 <irl> you may have noticed i did some triage on the tickets
15:10:09 <irl> made sure metrics-team was cc and such
15:10:26 <karsten> noticed that about cc, not sure about analysis tickets.
15:10:32 <irl> these two tickets were owned by arma
15:10:37 <karsten> I usually only see tickets that have metrics-team in cc.
15:11:04 <irl> i think we have two kinds of tickets in metrics/analysis
15:11:16 <irl> there are vague research ideas and then tasks that metrics can help with
15:11:37 <irl> i suggested to arma that we might split this component and make a new research component that might contain open questions
15:11:45 <karsten> sounds like a great plan.
15:11:50 <irl> or just close these tickets and maybe move the better open questions to the research portal
15:11:55 <irl> as trac isn't really the best place for them
15:12:03 <karsten> also an option, if we do that second step.
15:12:06 <irl> these are more likely to be answered by external researchers
15:12:19 <karsten> does the research portal exist?
15:12:28 <karsten> or what's the timeline there?
15:12:29 <irl> http://research.torproject.org/
15:12:32 <karsten> ahh.
15:12:41 <irl> so it exists as that
15:12:47 <irl> and there is an ideas page
15:12:49 <karsten> weren't there plans to make an actual research portal?
15:12:58 <karsten> that website is not updated very often.
15:13:03 <iwakeh> true.
15:13:07 <karsten> (read as: never)
15:13:14 <iwakeh> even more true
15:13:42 <irl> it's a static site, once the other portals are up i guess it's really easy to migrate it to lektor
15:14:00 <karsten> but, we could start with plan A (split components) and do plan B (move to research portal) as soon as that exists.
15:14:24 <irl> ok, where do you think is the best place to discuss this?
15:14:33 <irl> or do i just do it?
15:14:41 <karsten> where would that new component live?
15:14:51 <karsten> Core Tor?
15:14:51 <irl> "Research/Open Questions" probably
15:15:05 <irl> they're not core tor
15:15:31 <karsten> well, nobody will feel responsible for a new top-level component Research.
15:15:39 <karsten> unless somebody says they will.
15:16:32 <karsten> I'm not sure where to put it.
15:16:34 <irl> there is the same problem moving it to core tor though
15:16:46 <karsten> yes, we'd have to talk to the network team first in that case.
15:16:52 <karsten> maybe we should do that in any case.
15:16:55 <irl> i will check on the timeline for the research portal and see what exists there already
15:16:57 <irl> maybe there are plans
15:17:04 <karsten> also a good plan.
15:17:09 <iwakeh> and a sub-component of Metrics/Analytics ?
15:17:18 <iwakeh> fr sorting
15:17:20 <karsten> sub-sub?
15:17:31 <iwakeh> might be a trac limit.
15:17:31 <karsten> for sorting, just pick Metrics/Research ?
15:17:36 <irl> they're not really analysis at all
15:17:46 <iwakeh> hmm, this yields something different.
15:17:49 <irl> they're "it would be good if arma would write a paper on X" tickets
15:18:42 <karsten> so, do you need such a Metrics/Something component for the sorting?
15:18:46 <karsten> I can create one.
15:18:54 <karsten> (maybe you can, too.)
15:18:57 <irl> let's leave them where they are for now
15:19:03 <karsten> okay.
15:19:08 <irl> no point in moving them twice
15:19:16 <iwakeh> maybe tag?
15:19:37 <iwakeh> now, that you read through them already?
15:19:42 <irl> research-ideas would be a good tag
15:19:46 <irl> i can do this
15:19:59 <iwakeh> that'll safe time when moving is possible.
15:20:05 <irl> yep, good idea
15:20:05 <karsten> great!
15:20:21 <karsten> out of topics?
15:20:23 <karsten> :)
15:20:27 <irl> for real this time (:
15:20:34 <iwakeh> oh
15:20:36 <iwakeh> kay
15:20:40 <iwakeh> :)
15:20:41 <irl> heh
15:20:44 <karsten> by the way, let's skip next week's meeting.
15:20:48 <karsten> heh
15:20:53 <irl> ok cool
15:21:03 <iwakeh> ok
15:21:26 <karsten> that's all then!
15:21:37 <karsten> thanks, and bye, bye! :)
15:21:41 <iwakeh> bye, bye!
15:21:41 <irl> bye!
15:21:47 <karsten> #endmeeting