14:29:23 <karsten> #startmeeting metrics team
14:29:23 <MeetBot> Meeting started Fri Oct 20 14:29:23 2017 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:29:23 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:29:29 <karsten> hello!
14:29:45 <karsten> https://storm.torproject.org/shared/Ou-1QRctynWbF4yedi-MfDsjImFMFSIEP20fbVGCPRa <- agenda pad
14:29:54 <karsten> for today's metrics team meeting.
14:30:07 * karsten spotted iwakeh, but is irl around, too?
14:30:32 <iwakeh> On the pad?
14:30:46 <karsten> well, on the pad and here.
14:30:52 * irl is everywhere
14:30:55 <karsten> hehe
14:31:00 <karsten> perfect!
14:31:14 <karsten> I'll start with the first topic then:
14:31:17 <iwakeh> hi irl!
14:31:20 <karsten> * roadmap feedback from Montreal [karsten]
14:31:42 <karsten> so, as expected, I received a lot of ideas what else we could be doing.
14:31:55 <karsten> I did not receive many ideas we can remove from the current plan.
14:32:18 <karsten> some of the new ideas seem really important, so we'll have to decide which of the existing ones to drop ourselves.
14:32:36 <karsten> I'll go through the list that I started on the pad.
14:32:36 <iwakeh> is there a list of ideas in a wiki page or similar?
14:32:44 <iwakeh> ah ok.
14:32:50 <karsten> there are notes on the wiki.
14:32:57 <karsten> but I did not clean them up very much yet.
14:33:04 <karsten> so, #1 is:
14:33:05 <karsten> - Twitter bot/external communication
14:33:10 <irl> was the twitter bot really popular enough to be first on that list?
14:33:20 <karsten> that idea was really well received.
14:33:29 <irl> oh cool
14:33:41 <karsten> it's not necessarily the most important one, but there's certainly request for having this bot.
14:33:42 <iwakeh> what contents were expected for the bot?
14:34:06 <karsten> I think people started looking at @TorAtlas and liked that a lot.
14:34:19 <karsten> which doesn't mean that it's too late to discuss and decide ourselves what contents we want to provide.
14:34:24 <iwakeh> so, in the direction of our earlier mail thread.
14:34:34 <karsten> yes. I replied to that a few days ago.
14:34:38 <karsten> when you were afk.
14:34:49 <karsten> it's something we should include in the roadmap and pick up over the next weeks.
14:34:57 <iwakeh> true.
14:35:00 <karsten> and we should include steph and tommy.
14:35:08 <irl> so the bot basically has "templates" and then fills in data from onionoo
14:35:18 <irl> there's no reason we cant use the CSVs or OONI data or whatever
14:35:30 <karsten> sounds good.
14:35:42 <karsten> to be honest, I haven't even looked at the code yet,
14:35:47 <irl> those graphics are just Graphics2D things
14:35:56 <karsten> or put any thoughts into implementation questions.
14:36:04 <irl> #23799
14:36:25 <irl> i don't know who to poke for that
14:36:29 <iwakeh> metrics-bot?
14:36:41 <karsten> yes, metrics-bot might make more sense.
14:36:50 <karsten> poke for creating the repo?
14:36:54 <irl> yep
14:37:01 <irl> and i'm open to renaming it to metrics-bot
14:37:21 <karsten> fine question. I'm not sure who's creating repos these days. but I can find out.
14:37:25 <irl> i also see it being an irc bot that can do what zwiebelbot does for bugs but for relays
14:37:28 <iwakeh> cc karsten
14:37:36 <iwakeh> on the ticket.
14:37:48 <karsten> irl: sounds great!
14:37:51 <karsten> iwakeh: makes sense!
14:37:59 <irl> Cc:  karsten added
14:38:06 <karsten> alright.
14:38:13 <karsten> moving on?
14:38:19 <irl> yep
14:38:20 <iwakeh> fine.
14:38:26 <karsten> - OONI graphs
14:38:37 <karsten> this is also something that people want to see.
14:38:54 <karsten> also to show closer cooperation between metrics and ooni, which makes sense.
14:39:03 <iwakeh> on metrics.tp.o?
14:39:07 <karsten> what we discussed is that ooni will provide CSV files to us.
14:39:08 <karsten> yes.
14:39:19 <karsten> with tor-related stats.
14:39:21 <irl> ok, this sounds cool
14:39:29 <iwakeh> yes, indeed.
14:39:46 <karsten> for example, they have a vanilla tor test that tries to bootstrap from wherever there are ooni clients.
14:40:15 <karsten> the plan is to visualize how these connections are successful in different parts of the world, and maybe how fast that goes.
14:40:30 <karsten> and there's another data set about reachability of default bridges shipped with tor browser.
14:40:48 <irl> i remember they previously had a deliverable to explain the ooni pipeline, so we should get the detailed explanations of what is in the CSVs
14:40:50 <karsten> so, useful stuff. and we don't have to write the aggregation code. *phew*
14:41:04 <iwakeh> hehe.
14:41:22 <karsten> we'll certainly learn how they're making their CSVs.
14:41:30 <iwakeh> @irl yes that'll be important.
14:41:36 <karsten> I think we'll be part of the process to design them.
14:41:48 <karsten> so, something for the roadmap and for the upcoming weeks or months.
14:41:55 <irl> ok cool
14:42:05 <karsten> next?
14:42:10 <iwakeh> yes
14:42:14 <irl> yep
14:42:15 <karsten> - IPv6 graphs
14:42:23 <karsten> that's something that teor wants.
14:42:32 <karsten> he's working on integrating ipv6 better in tor.
14:42:52 <karsten> and he'd like to see graphs on how ipv6 adaption is right now, and how that changes after making some changes.
14:42:53 <iwakeh> the thread on metrics-team?
14:43:08 <karsten> maybe? I don't know exactly. (so many threads.)
14:43:21 * irl had planned to add ipv6 graphs to metrics-bot
14:43:31 * iwakeh sighs, about the lists.
14:43:32 <karsten> in any case, we won't have to add new data to the tor daemon for this.
14:43:41 <karsten> which would take ages.
14:43:52 <iwakeh> true.
14:43:52 <karsten> what we do need to do is write aggregation code.
14:44:08 <karsten> but the cool thing is that we can still throw in the entire archive when we're done.
14:44:17 <karsten> so we don't need to do it all this afternoon!
14:44:25 <karsten> it will still be useful in 4 weeks. or 8.
14:44:37 <iwakeh> is there a deadline?
14:44:55 <karsten> this work is also something that people care about and that show how metrics are important for the other parts of tor.
14:45:09 <karsten> not that I know of. maybe we should ask.
14:45:13 <karsten> good point.
14:45:18 <iwakeh> so, higher prio than the other topics we just talked about.
14:45:28 <karsten> roughly the same, I think.
14:45:31 <Sebastian> sorry, I'm creating repos
14:45:37 <Sebastian> but I haven't done it much lately
14:45:38 <karsten> thanks, Sebastian!
14:45:47 <karsten> ah.
14:45:56 <iwakeh> metrics-bot
14:45:59 <Sebastian> way more reliable pinging me on irc than trac unforutnately
14:46:09 <irl> yes, Sebastian can you rename the thing to metrics-bot not atlasbot?
14:46:38 <Sebastian> Is tomorrow ok?
14:46:48 <iwakeh> sure.
14:46:48 <irl> Sebastian: yes, i'm updating the ticket
14:46:49 <karsten> irl: maybe quickly update the ticket?
14:46:52 <karsten> great!
14:46:55 <Sebastian> renaming is annoying because you need to be careful and do manual stuff
14:46:59 <Sebastian> thanks
14:47:04 <Sebastian> and sorry for the wait
14:47:11 <iwakeh> thanks!
14:47:16 <irl> (:
14:47:27 <karsten> alright. that's all about ipv6. moving on?
14:47:36 <iwakeh> yep
14:47:37 <irl> karsten: can we quickly
14:47:45 <karsten> yes?
14:47:49 <irl> there was a bug about or addresses and reachability and things
14:48:01 <irl> and tor clients don't check reachability of ipv6 or ports
14:48:12 <irl> is it dirauths that test reachability themselves before including them?
14:48:14 <irl> and not all do?
14:48:20 <karsten> ah.
14:48:36 <karsten> you mean #21637?
14:48:48 <irl> yes
14:49:06 <karsten> if so, turns out that tor26 and gabelmoo don't include relays as running if they are not reachable via ipv4 _and_ ipv6.
14:49:18 <karsten> there's more on the ticket, and I sent something to the dirauth list.
14:49:34 <karsten> what do you want to know?
14:49:45 <karsten> and note that I'm not too deep into this.
14:50:06 <karsten> I mainly cared about finding out why my task-21637 onionoo branch produced the documents it produced.
14:50:19 <irl> i mostly just want to know what to do for atlas
14:50:20 <karsten> teor might know more.
14:50:36 <karsten> ah. nothing at this point.
14:50:36 <irl> if there's an ipv6 in or_addresses, is it reachable or untested?
14:50:53 <karsten> reachable.
14:50:57 <irl> ok
14:51:02 <irl> then this is fine
14:51:13 <karsten> with #21637 we're adding an unreachable_or_addresses field.
14:51:19 <irl> i couldn't answer this question yesterday on #tor, now i can
14:51:19 <karsten> with the ones that are not reachable.
14:51:23 <karsten> ah, cool!
14:51:28 <irl> we can move on
14:51:34 <karsten> - better tool support for relay operators
14:51:40 <karsten> so, relay operators are sad.
14:51:49 <irl> oops
14:51:52 <karsten> and we discussed a few ideas how to make them less sad.
14:51:58 <iwakeh> ok?
14:52:12 <karsten> the part that is relevant to us is that we can provide them with better tools.
14:52:24 <iwakeh> what kind?
14:52:28 <karsten> longer term thing could be reviving tor weather.
14:52:35 <irl> someone has already
14:52:50 <karsten> medium term thing would be an atlas with more gamification components.
14:53:07 <karsten> really? I heard about torweather.org, but I couldn't find a running instance.
14:53:09 <irl> https://github.com/thingless/torweather/blob/master/README.md
14:53:13 <irl> it's running
14:53:14 <karsten> I didn't look in detail, though.
14:53:15 <irl> i got mail from it
14:53:19 <karsten> ah, neat.
14:53:26 <karsten> should we add it to metrics.tp.o?
14:53:29 <karsten> as a link?
14:53:36 <irl> i mean, there's not really a link other than the sources
14:53:37 <iwakeh> ask the maintainer?
14:53:41 <irl> a lot was stripped out of it
14:53:47 <irl> to make it easier to run
14:54:01 <karsten> do we know who the maintainer is? (other than seeing their github account?)
14:54:11 <irl> nope
14:54:24 <karsten> what I mean: is there a website where I can sign up to get notifications?
14:54:31 <karsten> most relay operators won't care about the sources.
14:54:44 <irl> ah, no
14:55:08 <karsten> so, I need to run it myself to get notified about my own relays?
14:55:19 <irl> no, it'll scrape contactinfo
14:56:16 <iwakeh> so, it's run somewhere and notifies relay ops?
14:56:18 <karsten> hmmmm. so, if I were running a few relays, what would I do?
14:56:35 <irl> it just scrapes email addresses from contactinfo
14:56:41 <irl> when a relay is down, it'll tell you
14:56:43 <irl> by email
14:56:53 <irl> there is an unsubscribe option and it doesn't send more than one mail
14:56:54 <iwakeh> regardless of any subscription?
14:56:56 <irl> yes
14:57:01 <karsten> uhm.
14:57:03 <iwakeh> understood.
14:57:06 <irl> you can only unsubscribe, not subscribe
14:57:14 <iwakeh> not really great idea.
14:57:22 <karsten> so, this is just an example.
14:57:31 <irl> i do wonder if such notifications are in scope for metrics-bot
14:57:37 <karsten> we found that a working notification service might help relay operators.
14:57:53 <karsten> well, maybe for high-profile relays.
14:57:57 <iwakeh> yes, the bot might be a good place.
14:57:58 <irl> it could notify by email/tweet/irc/xmpp
14:58:18 <iwakeh> Metrics messaging engine ;-)
14:58:26 <karsten> ah, but directed to the relay operator only?
14:58:30 <irl> yep
14:58:33 <karsten> rather than shouting it out to the world.
14:58:39 <karsten> that would make sense then.
14:58:40 <karsten> :)
14:58:53 <karsten> but again, that's just an idea.
14:59:05 <karsten> what we should do is brainstorm how we can make the life of relay operators easier.
14:59:25 <karsten> okay, next one?
14:59:33 <irl> yep
14:59:41 <karsten> - PrivCount proposals
14:59:49 <karsten> we had a good meeting with the privcount folks.
15:00:00 <karsten> they finished one proposal and discussed it a few weeks ago.
15:00:09 <karsten> there will be more such proposals, some of which affecting us more than this one.
15:00:35 <karsten> for example, there will be one for including noise. and we'll have to know what's going on there, because we'll later have to handle that noise.
15:00:50 <karsten> or there will be another one for versioning statistics reported by privcount.
15:00:58 <karsten> like, how do we add or remove statistics.
15:01:05 <karsten> and two more or so.
15:01:23 <karsten> these will be written in the next months, and we should be part of that.
15:01:45 <karsten> good thing we already have an L task for this.
15:01:56 <karsten> see also the notes on the wiki about this session.
15:02:11 <iwakeh> were there other requests?
15:02:11 <iwakeh> about tools for r-ops?
15:02:13 <iwakeh> just downtime, t-shirt, ...?
15:02:19 <iwakeh> maybe, ask in the relay ml and collect ideas?
15:02:41 <iwakeh> wiki link?
15:02:49 <karsten> I'd say let's brainstorm a little, suggest something to relay operators, and ask for other ideas.
15:03:02 * karsten finds the link..
15:03:06 <iwakeh> fine.
15:03:18 <karsten> https://trac.torproject.org/projects/tor/wiki/org/meetings/2017Montreal/Notes/PrivCount
15:03:58 <karsten> keep in mind that these notes are relatively quickly written while talking about things, so nothing is written in stone yet.
15:04:19 <karsten> but I was under the impression that there's a common ground for making this work.
15:04:48 <iwakeh> fine, so we'll notice
15:05:04 <karsten> ah, and we have a plan for making the next proposal discussion work better despite timezones.
15:05:09 <iwakeh> via tor-dev or will there be other means of looking at proposals early?
15:05:26 <karsten> we might just use a pad and leave it open for a week where everyone can participate whenever they're awake.
15:05:41 <iwakeh> good :)
15:05:44 <karsten> I'm sure we'll notice somehow when new proposals are being written.
15:06:03 <iwakeh> ;)
15:06:20 <karsten> I have two dates here: feb 2018 and mid-2018. not sure which was which. but I figured neither of them is oct or nov 2017, so we're good.
15:06:27 <karsten> for now.
15:06:40 <karsten> alright.
15:06:44 <karsten> - inter-team dependencies
15:06:57 <karsten> we'll need to move forward with the roadmap draft until next thursday.
15:07:16 <karsten> the idea is that all teams have roadmaps until then to identify inter-team dependencies.
15:07:25 <iwakeh> another pad/googledoc meeting thursday?
15:07:50 <iwakeh> and, reviewing and adding the new topics earlier.
15:07:52 <karsten> regarding the tool, it might be that our google doc is sufficient.
15:08:18 <karsten> regarding the timing, I was planning to put in more ideas over the next days.
15:08:25 <karsten> including feedback above.
15:08:34 <iwakeh> good.
15:08:43 <karsten> I just had to write some code this week after talking about writing code for so long.
15:09:09 <karsten> alright. I think that's all for roadmap feedback.
15:09:25 <karsten> moving on to the next topic?
15:09:38 <iwakeh> I moved irl's topic as it is montreal related.
15:09:54 <karsten> * From the meeting: "Working bad: Tor Atlas backend is always down after a few searchers." ??
15:09:57 <karsten> you mean this one?
15:10:00 <iwakeh> oh, unless there is another feedback topic.
15:10:07 <iwakeh> yes
15:10:08 <karsten> not that I know.
15:10:22 <karsten> do we know anything about that issue?
15:10:33 <irl> who reported it when where?
15:10:41 <karsten> somebody in montreal.
15:10:54 <iwakeh> anonymous person?
15:11:00 <karsten> we had a session where we wrote "works better", "works the same", or "works less well" on stickies.
15:11:09 <karsten> and this one was on a "works less well" sticky.
15:11:10 <iwakeh> ah.
15:11:22 <karsten> (we had a few "works better" stickies, too. :) )
15:11:28 <irl> i've not seen this
15:11:50 <karsten> I also had atlas hanging/not making progress earlier today.
15:12:00 <karsten> when I was looking into the ipv6 thing.
15:12:02 <irl> i've seen that we don't really get much backend being down since the new onionoo servers came along
15:12:25 <karsten> could it be the second real backend?
15:12:29 <karsten> or the load balancing?
15:12:40 <karsten> we have lots of new caches.
15:12:43 <irl> i just can't reproduce it
15:12:57 <iwakeh> maybe, that was an older notion?
15:12:57 <irl> i get more failures from the twitter api than i do from onionoo
15:12:58 <karsten> but one that is a cache and separate backend.
15:13:08 <karsten> okay.
15:13:20 <karsten> I think most things on these stickies were quite recent.
15:13:29 <karsten> but, anyway,
15:13:32 <irl> maybe it's when i'm asleep?
15:13:36 <karsten> maybe watch out for any issues.
15:13:37 <karsten> hehe
15:13:40 <iwakeh> :-)
15:13:53 <karsten> we can move on.
15:14:09 <iwakeh> sure.
15:14:14 <karsten> * next steps for Sponsor13 deliverables [iwakeh] (if time permits)
15:14:28 <iwakeh> right, I thought about topics.
15:14:34 <karsten> ok.
15:14:38 <iwakeh> and how to approach the doc structure.
15:14:43 <iwakeh> another idea is
15:14:54 <iwakeh> to collect topics/questions raised in
15:15:06 <iwakeh> mailing lists (as happened recently)
15:15:15 <iwakeh> on a wiki page in order
15:15:31 <iwakeh> to keep track and really address these in the end.
15:15:36 <karsten> collecting ideas on a wiki page sounds great to me!
15:15:44 <karsten> right now I'm collecting them, erm, in my inbox.
15:15:47 <iwakeh> ok, I'll start one.
15:15:51 <karsten> please do!
15:16:10 <karsten> maybe also put your ideas about the document structure there.
15:16:14 <karsten> and topics.
15:16:16 <iwakeh> yes.
15:16:40 <karsten> I'd say it's perfectly fine to collect ideas for the next few weeks or even until the end of 2017.
15:16:50 <iwakeh> ok.
15:16:51 <karsten> while we work off things that seem more urgent.
15:17:01 <iwakeh> next topic?
15:17:06 <karsten> and start actually writing docs in early 2018.
15:17:08 <karsten> yep.
15:17:12 <karsten> * next steps webserver logs (CollecTor&metrics-lib) [iwakeh] (if time permits)
15:17:19 <karsten> all reviewed! :)
15:17:29 <iwakeh> I noticed.
15:17:31 <iwakeh> :-)
15:17:45 <karsten> though not exactly on the plane. anyway. ;)
15:17:49 <iwakeh> the replies will be high prio.
15:18:02 <iwakeh> for both.
15:18:06 <karsten> okay! I'll prioritize responding and re-reviewing.
15:18:07 <irl> just a thing from #tor, someone found the onion service for the sanitised logs and thought they'd stumbled onto a secret hacking operation
15:18:24 <iwakeh> oh
15:18:43 <irl> maybe an index.html on the virtualhost to explain what is there and a link to the out folder would help
15:18:43 <karsten> the onion service for https://webstats.torproject.org/ ?
15:18:45 <irl> yep
15:18:58 <karsten> we'll soon shut that down.
15:19:01 <iwakeh> this will go away.
15:19:06 <irl> ah ok
15:19:07 <karsten> where soon means in the next couple of weeks.
15:19:12 <iwakeh> it'll be on collector.
15:19:16 <irl> cool (:
15:19:40 <karsten> okay. anything else on webstats?
15:19:47 <iwakeh> nope.
15:19:52 <karsten> last topic:
15:19:56 <karsten> * other priorities regarding tickets [all] (if time permits, this might also depend on the first topic)
15:20:01 <karsten> heh
15:20:06 <karsten> fine question.
15:20:11 <iwakeh> roadmap
15:20:16 <iwakeh> for all of us.
15:20:33 <karsten> you mean prioritize the roadmap over tickets?
15:20:50 <iwakeh> not really, but not drop it.
15:21:07 * iwakeh wants to know if there are pending reviews open.
15:21:19 <karsten> lots.
15:21:31 <karsten> but, that doesn't mean they are higher prio than the roadmap.
15:21:42 <karsten> the roadmap has a deadline.
15:21:43 <iwakeh> ok.
15:22:06 <karsten> so, from all the tickets in needs_review, I'm not sure if there's a clear order regarding priority.
15:22:18 <iwakeh> that's fine,
15:22:22 <karsten> if you want to review something, maybe pick something that looks easy.
15:22:27 <iwakeh> I'll define it myself :-)
15:22:34 <karsten> sounds good to me. :)
15:22:49 <iwakeh> just wanted to make sure nothing is overlooked.
15:22:56 <karsten> nothing is on fire.
15:23:03 <iwakeh> good!
15:23:14 <karsten> alright. I think we're done then.
15:23:18 <irl> yay
15:23:21 <iwakeh> yep.
15:23:28 <karsten> good meeting!
15:23:35 <irl> (:
15:23:35 <iwakeh> yes
15:23:41 <karsten> I'll ping you via email as soon as I have something new for the roadmap.
15:23:46 <iwakeh> fine.
15:24:01 <karsten> have a nice weekend, and let's talk more next week!
15:24:09 <iwakeh> you too. bye, bye!
15:24:11 <irl> i'll rename atlasbot, and will email when sources are uploaded
15:24:17 <irl> though i won't be upset if you don't look immediately
15:24:20 <iwakeh> great!
15:24:21 <karsten> sounds good, irl!
15:24:27 <karsten> bye, bye! :)
15:24:33 * irl -> coffee
15:24:35 <karsten> #endmeeting