15:12:58 <karsten> #startmeeting metrics team meeting
15:12:58 <MeetBot> Meeting started Thu Feb  6 15:12:58 2020 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:12:58 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:13:07 <karsten> now I did!
15:13:10 <karsten> thanks. :)
15:13:14 <dennis_jackson> Ah okay - I don't think I've seen the proposal, just the roadmap where it looks like it somewhat fits
15:13:19 <irl> i think it's fine to say it's out of scope, but we should see what mikeperry says and if it needs to be in-scope somewhere else
15:13:39 <karsten> which roadmap?
15:13:56 <dennis_jackson> https://pad.riseup.net/p/0K2Q5uuN0CbjbOGxJQsL-keep
15:14:42 <karsten> we kinda stopped making the roadmap at that objective.
15:14:58 <karsten> I think feb 10 is when we're continuing.
15:15:03 <dennis_jackson> Ah okay
15:15:08 <karsten> so, this will be an important question for that meeting.
15:15:24 <gaba> :)
15:15:31 <dennis_jackson> I have a draft email on those graphs which I will try to fire off earlier rather than later then :)
15:15:40 <karsten> which ones?
15:15:47 <karsten> the ones on #33076?
15:15:53 <dennis_jackson> Yes
15:15:59 <karsten> sounds great!
15:16:00 <gaba> dennis_jackson: this is a project that we sent a proposal for funding but still have not been funded. we are going to be working on it soon
15:16:37 <karsten> looking forward to reading that email!
15:16:39 <dennis_jackson> Great! Fingers crossed then
15:17:01 <karsten> okay, moving on?
15:17:12 <dennis_jackson> :+1:
15:17:27 <karsten> Torflow/sbws transition (#33077) (karsten)
15:17:41 <karsten> I started working on that and realized that metrics-lib is missing a few pieces.
15:18:03 <karsten> and when I worked on extending metrics-lib, I ran into several smaller bugs that went unnoticed so far.
15:18:15 <gaba> oh
15:18:18 <karsten> did you know that you shouldn't invoke overridable methods from constructors? ...
15:18:31 <karsten> that kept me busy for too long.
15:18:32 <irl> that sounds safe enough
15:18:43 <karsten> not doing it is safe.
15:18:52 <karsten> we're doing it.
15:19:12 <karsten> anyway, I hope to have an update on #33077 by tomorrow.
15:19:18 <irl> ok cool
15:19:38 <karsten> this was mostly an update/rant.
15:19:43 <karsten> moving on?
15:19:47 <irl> ok
15:20:02 <karsten> Onionoo 8.0 announcement (karsten)
15:20:19 <karsten> we're going to have a new major version of the Onionoo protocol in ~2 weeks.
15:20:38 <karsten> in the past we announced these updates. we didn't do this yet this time.
15:20:42 <gaba> ok
15:20:48 <karsten> is it sufficient to include something in the meeting notes?
15:21:02 <gaba> mmm. I think it would be better to do an announcement for major versions
15:21:15 <karsten> ok. where?
15:21:57 <karsten> metrics-team@?
15:22:30 <irl> tor-project@ maybe
15:22:38 <karsten> works for me.
15:22:42 <gaba> yes
15:22:48 <irl> with the reason that this is also reminding people that onionoo exists even if they are not users yet
15:23:16 <karsten> sounds good.
15:23:32 <karsten> next topic?
15:23:39 <irl> ok
15:23:50 <karsten> Exit scanner timeline (irl)
15:24:01 <irl> https://people.torproject.org/~irl/exitscan.html
15:24:18 <irl> i expect to have 3, 4, 5 and 6 done this week or monday evening at the latest
15:24:43 <irl> task 7 is dependent on how quickly i can do the thing i've never done before
15:24:59 <irl> which is hard to estimate but i'm hoping it's around a week
15:25:01 <karsten> is mid-feb still realistic?
15:25:19 <irl> plus or minus a few days, i think so
15:25:23 <karsten> okay.
15:26:00 <gaba> ok
15:26:15 <karsten> moving on?
15:26:19 <irl> ok
15:26:33 <karsten> Announcement: Simply Secure starting to work on metrics portal (user research and design). They will be interviewing people for their user research.
15:26:38 <gaba> yep, only that
15:26:42 <irl> woo
15:26:50 <karsten> I saw the kick-off notes. interesting stuff!
15:27:00 <gaba> estimated date for their work to end is August
15:27:06 <irl> i did not look yet, but i will try to do that this week too
15:27:11 <irl> ok cool
15:27:56 <karsten> okay, if it's just the announcement, moving on?
15:28:17 <gaba> yes
15:28:17 <irl> yes
15:28:19 <dennis_jackson> Sorry - is this a metrics portal for user research?
15:28:31 <dennis_jackson> Or a new metrics portal, for which they want to ask users of the current one?
15:28:39 <gaba> dennis_jackson: this is a user research for the new metrics data website
15:28:51 <gaba> i can tell them to interview you too about how to use that data :)
15:28:56 <dennis_jackson> Ah, fantastic :)
15:29:05 <dennis_jackson> Always happy to share my 2 cents
15:29:22 <gaba> They will be coming to this meetings sometimes to share the status of the project
15:29:48 <dennis_jackson> great
15:30:26 <karsten> moving on?
15:30:33 <gaba> yes
15:30:36 <dennis_jackson> yes
15:30:36 <irl> ok
15:30:38 <karsten> Anything to bring to the network health team's plate for the next 3 months? (thinking about onionperf & scalability)
15:30:51 <karsten> is there plate not full enough yet? :)
15:30:55 <karsten> their*
15:31:03 <gaba> hehe
15:31:26 <gaba> well, it is moslty related to the onionperf work that you are going to do. Just checking if there is anything
15:32:15 <gaba> we can continue and if there is anything we can talk on monday
15:32:18 <gaba> when we talk about that project
15:32:29 <karsten> I noticed that GeKo is looking at various tickets as possible candidates.
15:32:46 <karsten> it seems to me they have a lot of things already they can work on.
15:32:51 <gaba> yes, this is the pad where we are working on his roadmap: https://pad.riseup.net/p/network-health-team-2020Q1-roadmap-temporal
15:32:54 <gaba> yes
15:33:30 <karsten> okay, let's talk more about this on monday.
15:33:36 <karsten> (unless others have ideas.)
15:33:41 <irl> i think we should work out which things are metrics and which things are network health generally
15:33:46 <irl> but right now i don't have specific ideas
15:33:57 <gaba> yes
15:33:58 <gaba> ok
15:34:03 <irl> we should avoid duplication of effort
15:34:09 <karsten> agreed.
15:34:40 <karsten> moving on?
15:35:22 <gaba> ok
15:35:23 <mikeperry> oh hey I just saw the scrollback
15:35:29 <karsten> hi mikeperry!
15:35:35 <gaba> hi!
15:35:52 <mikeperry> why do we think these graphs are out of scope?
15:36:08 <mikeperry> checking ticket
15:36:12 <mikeperry> we need that for sbws eval
15:36:17 <mikeperry> should we just stick with torflow?
15:36:26 <karsten> for O3.2 we had to throw out everything that's not based on onionperf.
15:36:48 <karsten> well, the torflow/sbws transition can happen regardless. we can make graphs ad-hoc.
15:36:51 <mikeperry> I just need tooling that produces the graphs. it doesn't have to be on metrics website
15:37:00 <mikeperry> like scripts that I can run
15:37:17 <karsten> and the thing we build for MOSS will likely not be on the metrics website.
15:37:32 <karsten> it will be scripts. well-documented scripts, but scripts.
15:37:56 <karsten> I can make graphs for the torflow/sbws transition as needed.
15:38:09 <mikeperry> that is ok. so long as they are documented and I and other researchers can just pull down data from somewhere and run and re-run them, that is good enough
15:38:55 <karsten> the scripts to make CDF-TTFB and CDF-DL graphs will be documented and generally usable for devs.
15:38:56 <mikeperry> the descriptor archives, for example, are fine
15:40:01 <karsten> but for CDF-Relay-* graphs we'd have to process more descriptor types, include them in the database schema, and so on. it's just more work, and we had to cut down the scope for budget reasons.
15:40:36 <karsten> should we move this to #33076?
15:40:43 <mikeperry> but there will still be documented steps for those scripts right?
15:41:03 <karsten> for CDF-Relay-* graphs?
15:41:08 <mikeperry> for the CDF-Relay-* graphs? so I can re-run them if I need?
15:41:37 <karsten> I can clean up my quick-and-dirty code a little. but it will be java and R, rather than python/matplotlib.
15:41:52 <dennis_jackson> I would also like to play around with the code if possible
15:42:18 <karsten> okay, I can revive the metrics-tasks git repository for this.
15:42:44 <dennis_jackson> That would be great
15:43:28 <mikeperry> yeah. and some docs for setup + which datasources/how it works with the data
15:44:17 <karsten> ok. :)
15:44:47 <karsten> by the way, this shouldn't stop the brainstorming about possible graphs we might want for scalability/performance experiments.
15:44:49 <mikeperry> particularly I will need to know how to change the time ranges, and hints at how I might filter out certain relays, or just examine only certain relays
15:45:54 <karsten> we should probably also try to get this funded to get it included in the real tool.
15:46:49 <karsten> continue on the tickets?
15:46:52 <mikeperry> ok
15:46:53 <gaba> ok
15:46:57 <karsten> cool!
15:47:03 <karsten> Needs review tickets: https://trac.torproject.org/projects/tor/query?status=needs_review&component=%5EMetrics&col=id&col=summary&col=component&col=owner&col=type&col=priority&col=milestone&col=reviewer&order=priority
15:47:39 <karsten> irl: nothing is super urgent there.
15:47:50 <karsten> if you prefer to continue exit scanner work, please so that.
15:47:51 <irl> is anything more important than the exit scanner?
15:47:51 <karsten> do*
15:47:56 <irl> right ok
15:48:21 <karsten> Defect tickets: https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=needs_information&status=needs_revision&status=new&status=reopened&type=defect&component=%5EMetrics&group=priority&col=id&col=summary&col=component&col=status&col=priority&col=changetime&desc=1&order=component
15:48:23 <gaba> yes, exit scanner is priority
15:49:15 <karsten> I'd rather focus on #33077 first and the defects I just found in metrics-lib.
15:49:40 <gaba> yes
15:49:41 <gaba> agree
15:49:49 <karsten> so, from my perspective, we can leave these defects where they are for the moment.
15:50:05 <irl> ok agreed
15:50:18 <karsten> last topic:
15:50:20 <karsten> Roadmap (gaba) https://trello.com/b/Mu5fYg53/tor-metrics-roadmap
15:51:07 <irl> i had added #33076 when it was filed, but i think that is the same as the one that was in review already
15:51:18 <gaba> yes
15:51:29 <gaba> focus on revieweing 33076 :)
15:51:49 <karsten> and revising as needed.
15:53:03 <karsten> I'll also leave #32978 there as a permanent reminder. not that I had a plan how to move that forward.
15:53:14 <irl> ok
15:53:58 <karsten> any more changes?
15:54:18 <gaba> ok
15:55:02 <karsten> are we good with the roadmap?
15:55:24 <gaba> it seems ok for now
15:55:37 <gaba> monday we are talking onionperf, right?
15:55:51 <karsten> onionperf and roadmap?
15:55:59 <gaba> 15UTC
15:56:08 <gaba> I sent you all an email earlier today about it
15:56:09 <gaba> yes
15:56:10 <gaba> roadmap
15:56:15 <karsten> sounds good!
15:57:12 <irl> ok
15:57:14 <karsten> okay, we're out of topics then.
15:57:46 <irl> cool
15:57:55 <karsten> next meeting next week? (and the roadmap/onionperf thing next monday.)
15:58:11 <irl> yes
15:58:27 <karsten> talk to you then! thanks, everybody. have a great rest of the week! bye. o/
15:58:30 <irl> bye!
15:58:39 <dennis_jackson> o/
15:58:42 <karsten> #endmeeting