14:32:31 <karsten> #startmeeting metrics team
14:32:31 <MeetBot> Meeting started Thu Jan 25 14:32:31 2018 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:32:31 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:32:39 <iwakeh> hi hiro!
14:33:04 <karsten> https://storm.torproject.org/shared/Oh4g0hNenh65QZNWRsIe5zxpB3e0axASeSgo5hKOp2A <- agenda pad
14:33:09 <karsten> irl2: the other meeting ended on time?
14:33:21 <irl2> karsten: yes, it was shorter than expected
14:33:26 <karsten> nice. :)
14:33:52 <karsten> agenda pad already has quite a few topics.
14:34:04 <iwakeh> only a few ;-)
14:34:05 <irl2> many many topics
14:34:05 <karsten> anything that is missing?
14:34:18 <karsten> (to be fair, I pasted the entire roadmap into it, including comments.)
14:34:42 <iwakeh> (from where?)
14:34:45 <karsten> but if nothing is missing, I'd say let's start.
14:34:50 <karsten> from trac, reformatted.
14:34:55 <iwakeh> ok
14:34:57 <karsten> I'll add updates back.
14:35:25 * iwakeh waiting for storm to load and reload and load ...
14:35:29 <karsten> ugh
14:36:03 <karsten> we can also start with another topic that doesn't require the pad.
14:36:05 <karsten> let's do that.
14:36:08 <karsten> * Compass shut down (irl)
14:36:18 <irl2> this is just an announcement
14:36:19 <karsten> what remains to be done there?
14:36:21 <karsten> aha!
14:36:23 <irl2> that compass is gone
14:36:30 <karsten> indeed.
14:36:42 <irl2> that's now two services gone, the first being globe
14:36:43 * karsten makes a note for the monthly report.
14:36:46 <irl2> making it all more managable
14:36:51 <karsten> yep.
14:37:25 <karsten> okay, the next one is:
14:37:27 <karsten> * onionperf errors (#24996) (irl)
14:37:40 <karsten> there's already a patch on the ticket.
14:38:06 <karsten> though I think you added the topic prior to that.
14:38:30 <irl2> i think i filed the ticket on behalf of armadev
14:38:40 <karsten> sounds good.
14:38:42 <irl2> and i think i added this to the agenda because hiro
14:39:10 <karsten> well, there are indeed a lot of failures and timeouts.
14:39:11 <hiro> I just wanted to make sure there was not something wrong with the onionperf vms
14:39:13 <karsten> but not 200%.
14:39:23 <karsten> maybe there is something wrong.
14:39:23 <irl2> but the comment looks sane on the ticket
14:39:26 <hiro> january there have been various updates
14:39:42 <karsten> irl2: want to leave a comment (now or after the meeting)?
14:39:42 <hiro> because of meltdown and spectre and those have been restarted a few times
14:39:48 <karsten> aha!
14:39:52 <iwakeh> oh
14:40:05 <karsten> still, that would be temporary downtimes.
14:40:09 <irl2> karsten: i will try to comment this evening on the patch
14:40:13 <hiro> so yesterday when irl was chatting about it I mentioned this
14:40:17 <karsten> irl2: thanks!
14:40:48 <hiro> yes
14:40:52 <karsten> okay. I didn't look in detail, but I'm under the impression that onionperfs don't provide very reliable measurements these days.
14:41:00 <hiro> also I think only hk is affected more than other vms
14:41:13 <karsten> how about we look again once the graph is fixed?
14:41:18 <hiro> sure
14:41:25 <iwakeh> makes sense.
14:41:44 <irl2> ok
14:41:46 * iwakeh has the pad again :-)
14:41:51 <karsten> great!
14:42:03 <karsten> should we start with reviewing the roadmap then?
14:42:10 <iwakeh> yep
14:42:22 <karsten> - Update all code to use Java 8 features (M) (Short-term goals (Q4/2017), Code)
14:42:36 <karsten> I said we started this, and it's ongoing.
14:42:39 <karsten> anything to add?
14:42:51 <iwakeh> reflects reality.
14:42:57 <karsten> ok.
14:43:02 <karsten> - Switch ExoneraTor and metrics-web from Tomcat to embedded Jetty, and harmonize directory structures (M) (Short-term goals (Q4/2017), Code)
14:43:07 <karsten> this is done.
14:43:16 <iwakeh> done.
14:43:20 <karsten> - Deploy better notification system for operational issues (M) (Short-term goals (Q4/2017), Operation)
14:43:23 <karsten> - discussed, not implemented anything yet
14:43:33 <karsten> at some point we should continue with that.
14:43:35 <irl2> onionoo is sort of monitored by metrics-bot
14:43:37 <karsten> but probably not in january.
14:43:47 <karsten> oh?
14:43:54 <irl2> it tells me when it's broken in #tor-test which isn't really the best set up
14:44:04 <irl2> but as it's polling onionoo regularly, i figured an alert was easy to do
14:44:22 <karsten> ok.
14:44:50 <karsten> so, we can still do a lot more there.
14:44:54 <irl2> indeed
14:44:56 <karsten> also including other services.
14:45:10 <karsten> - Make sure that each service has at least two operators (M) (Short-term goals (Q4/2017), Operation)
14:45:18 <karsten> - CollecTor and Onionoo are done, ExoneraTor and metrics-web remain, Relay Search doesn't have moving pieces on the server
14:45:26 <iwakeh> yes, or setting up other nagios emails for notification?
14:45:27 <karsten> which services did I miss?
14:45:42 <iwakeh> i.e. CollecTor's space alert.
14:45:51 <iwakeh> for corsicum
14:45:52 <karsten> iwakeh: yes, using more nagios notifications would work, too.
14:46:47 <karsten> more work remains for notifications.
14:46:56 <karsten> regarding two operators,
14:47:15 <karsten> that is also something where we could do more.
14:47:27 <karsten> not in january, though.
14:47:48 <karsten> okay, moving on
14:47:52 <karsten> - Add web server logs as new data source (M) (Short-term goals (Q4/2017), Sources)
14:47:55 <karsten> - currently in progress
14:48:05 <iwakeh> should we mark the 'not in Jan' still in January etc. remarks?
14:48:11 <karsten> sure.
14:49:11 <karsten> - Provide metrics timeline events as both a table on Tor Metrics pages and as an RSS/Atom feed that is also syndicated via Twitter to increase community engagement (M) (Short-term goals (Q4/2017), News)
14:49:15 <karsten> - Tor Metrics pages done, RSS/Atom remains
14:49:27 <karsten> so, we did 50% of this, but I don't have plans for the remaining 50%.
14:49:36 <irl2> the RSS/Atom is just a JSP from the same data source
14:49:53 <irl2> i can do this but probably end of feb/march
14:50:11 <karsten> sounds good to me.
14:50:46 <karsten> (thanks!)
14:50:46 <karsten> - Make Atlas part of the Tor Metrics website (M) (Short-term goals (Q4/2017), Services)
14:50:50 <karsten> - half-way through, need to fully integrate it
14:51:04 <irl2> this is just needing time
14:51:06 <karsten> what was the schedule for that again?
14:51:23 <irl2> probably end of feb/march again
14:51:34 <karsten> great!
14:51:48 <karsten> - Make Compass part of Atlas in order to provide aggregated search results and details pages on Atlas and to finally shut down Compass (M) (Short-term goals (Q4/2017), Services)
14:51:52 <karsten> - done, right? Yes!
14:51:54 <karsten> yay. :)
14:51:56 <irl2> (:
14:52:03 <karsten> - Resolve ExoneraTor database slowness (M) (Short-term goals (Q4/2017), Services)
14:52:06 <karsten> - started by dropping requests for same day, database improvements remain
14:52:18 <karsten> I checked a week or so ago, and the slowness still exists.
14:52:34 <iwakeh> let's look into this more.
14:52:43 <karsten> but I don't have time to look deeper into this right now.
14:52:49 <karsten> it's a medium-sized project.
14:53:03 <iwakeh> I could after the webstats modules?
14:53:38 <iwakeh> depending on urgency of other tasks.
14:53:39 <karsten> hmm, or maybe we should make more progress on Sponsor 13 then?
14:53:45 <karsten> yeah, not sure.
14:53:56 <karsten> let's see when we're there, ok?
14:54:01 <iwakeh> keep it on the decision list.
14:54:05 <karsten> yep.
14:54:31 <karsten> - Add OONI graphs to Tor Metrics based on pre-aggregated data from vanilla Tor test and bridge reachability test (M) (Short-term goals (Q4/2017), Visualizations)
14:54:34 <karsten> - irl started this, needs more work, irl will meet with OONI people at Rome meeting to follow up
14:54:37 <karsten> sounds good.
14:54:46 <karsten> - Add IPv6 relay graphs to Tor Metrics based on existing descriptor contents (M) (Short-term goals (Q4/2017), Visualizations)
14:54:49 <karsten> - done; the remaining specification part belongs into Sponsor 13
14:54:55 <karsten> yay, something is done!
14:54:59 <iwakeh> :-)
14:55:10 <karsten> - Rename Tor Metrics components to make them easier to understand for people outside of the team (M) (Short-term goals (Q4/2017), About)
14:55:13 <karsten> - in progress (Atlas became Relay Search)
14:55:37 <irl2> should we do focus groups and similar things for this one?
14:55:44 <irl2> i'm unsure what level we take this too
14:55:46 <iwakeh> yes, and new modules adhere to the new structure.
14:56:03 <karsten> focus groups?
14:56:13 <irl2> get people outside the team together and try out names on them
14:56:18 <karsten> iwakeh: you mean package structure?
14:56:30 <iwakeh> yes
14:56:38 <karsten> irl2: yes, maybe!
14:56:49 <karsten> want to describe that briefly on the pad?
14:56:54 <irl2> ok
14:57:39 <karsten> - Document aggregated statistics files better (Sponsor 13) (L) (Medium-term goals (until Q3/2018), Research)
14:57:42 <iwakeh> package structure can be slipped in, like immediately with the webstats changes for collector.
14:57:42 <karsten> - started, but no more than 10% done
14:58:08 <karsten> iwakeh: yes. but we should also consider renaming existing packages when there are no pending merges.
14:58:38 <karsten> iwakeh: in fact, should we list all unmerged branches and see which packages we can just rename now?
14:58:38 <iwakeh> yes, that's what I mean after the merge of the webstats module.
14:59:08 <iwakeh> could work.
14:59:26 <karsten> ok.
14:59:38 <karsten> - Put external research data on the website, including Rob's phantomtrain OnionPerf data and Yixin's BGP monitoring data (M) (Medium-term goals (until Q3/2018), Research)
14:59:41 <karsten> - iwakeh started this
14:59:48 <karsten> this is low-priority, I think.
15:00:14 <iwakeh> nothing urgent. but on my list.
15:00:19 <karsten> ok.
15:00:23 <karsten> - Re-process bridge descriptor tarballs to retain “contact” lines in server descriptors and “fingerprint” lines in statuses (M) (Medium-term goals (until Q3/2018), Sources)
15:00:27 <karsten> - no progress
15:00:38 <karsten> I just need to start this thread on tor-dev@.
15:00:53 <karsten> but, not in january.
15:01:10 <karsten> - Support PrivCount development and integrate its data as successor to statistics reported by relays and bridges (L) (Medium-term goals (until Q3/2018), Sources)
15:01:13 <karsten> - not much to do for us yet
15:01:15 <karsten> is that correct?
15:01:25 <iwakeh> well, reading specs and watching.
15:01:36 <iwakeh> but no need to interfere in any way.
15:01:38 <karsten> - Help the Core Tor and Tor Browser teams with ongoing measurements (M) (Medium-term goals (until Q3/2018), Sources)
15:01:41 <karsten> - no progress (rather not much to do?) (not sure, weren't we talking about measuring something for the Tor Browser team? looking... was that #24018 ?)
15:01:44 <karsten> iwakeh: true.
15:02:18 <irl2> i remember we wanted to put tor browser into onionperf at some point
15:02:36 <karsten> uhm.
15:02:47 <iwakeh> now that you mention it, I vaguely remember
15:03:02 <irl2> using selenium or something similar
15:03:19 <irl2> but i also wonder if this task is just to ensure we have time in case measurements or analysis were requested
15:03:24 <iwakeh> comment:10 on the ticket
15:03:51 <karsten> okay. I didn't follow that ticket very much.
15:03:58 <karsten> so, more work remains.
15:04:14 <karsten> should we go through all roadmap items and assign a percentage done?
15:04:30 <karsten> just so that we have a rough idea what % of work is done after 4/12 months?
15:04:34 <iwakeh> maybe, next time?
15:04:40 <karsten> okay.
15:04:54 <iwakeh> today we become aware of all tasks and set the % next time.
15:05:25 <karsten> yes, makes sense.
15:05:30 <karsten> thanks for stopping me here. ;)
15:05:39 <karsten> alright!
15:05:48 <karsten> anything else on the roadmap topic?
15:06:15 <iwakeh> seems we addressed all topics.
15:06:34 <karsten> regarding the roadmap, yes.
15:06:37 <karsten> okay.
15:06:41 <karsten> * Rome meeting plans (karsten)
15:07:14 <karsten> iwakeh: IIUC, you're not going to make it to rome, right?
15:07:26 <karsten> :(
15:07:32 <iwakeh> correct
15:07:46 <karsten> should we try to schedule a pre-meeting with just the team?
15:07:50 <karsten> like last time?
15:08:23 <irl2> how far in advance would this be?
15:08:24 <iwakeh> depends
15:08:45 <karsten> I don't know, the idea is quite new.
15:08:56 <karsten> maybe second half of February?
15:09:25 <irl2> this is probably ok for me, i'm travelling at the start of february
15:09:43 <irl2> i have i think a paper deadline in february, so i'd need to check my calendar really
15:09:54 <karsten> okay. discuss more via email?
15:09:56 <iwakeh> Would April also be an option to discuss and plan with the march results?
15:10:07 <karsten> also worth thinking about, yes.
15:10:19 <irl2> that would perhaps work for me more easily
15:10:23 <iwakeh> mail is fine.
15:10:34 <karsten> great!
15:10:49 <irl2> for rome itself, did we decide a plan?
15:11:04 <irl2> karsten: you said don't book travel until we plan
15:11:11 <karsten> right.
15:11:29 <karsten> I could imagine being there on the team day anyway.
15:12:40 <karsten> we wouldn't be able to make plans for the whole team.
15:12:56 <karsten> but we could do some work while we're there.
15:13:04 <irl2> yes
15:13:23 <karsten> and maybe schedule meetings with other teams about specific topics.
15:13:34 <irl2> this sounds good
15:13:50 <karsten> okay, so, let's plan to be there for the team day then.
15:14:12 <irl2> is that "day 0" or is there a "day -1"
15:14:48 <karsten> depends on what day 1 is.
15:14:59 <irl2> heh, ok, i will mail you before booking anything
15:15:06 <irl2> to make sure we're both agreeing on what date is when
15:15:06 <karsten> great!
15:15:11 <karsten> yes, good idea.
15:15:22 <karsten> moving on?
15:15:26 <irl2> yes
15:15:30 <karsten> * if time permits: discuss: format for sponsor13 deliverables; decide on level of detail; (iwakeh)
15:15:42 <iwakeh> Maybe it is too much to discuss this now, but while reviewing I noticed
15:15:52 <iwakeh> that it would be helpful to know the future format and
15:16:06 <iwakeh> for example, explanations
15:16:18 <iwakeh> should be accompanied by formulae and
15:16:31 <iwakeh> concrete definitions.  I made more suggestion
15:16:44 <iwakeh> on the branch I supplied earlier.
15:16:48 <karsten> can you add all those thoughts and suggestions to the .md file?
15:16:51 <karsten> ah, you already did?
15:17:01 <iwakeh> yes, including
15:17:05 <karsten> I didn't read all emails before this meeting.
15:17:10 <iwakeh> an Ant task for generating it.
15:17:21 <karsten> I think it's the perfect time to discuss this now. as in, these days, not now at this very moment.
15:17:32 <karsten> great!
15:17:52 <karsten> I don't know what the right level of detail is.
15:17:54 <iwakeh> html would allow using mathjax
15:18:08 <karsten> I'm not sold on .md.
15:18:20 <iwakeh> md equals to html
15:18:23 <iwakeh> more or less
15:18:31 <iwakeh> you can use html inside md
15:18:41 <karsten> oh.
15:18:45 <iwakeh> I made some comments inside as example.
15:18:50 <karsten> sounds good.
15:18:54 <iwakeh> in the branch wait ..
15:19:13 <iwakeh> #24217
15:19:15 <karsten> maybe this is a good question for the people I listed at the top of the document.
15:19:28 <iwakeh> https://gitweb.torproject.org/user/iwakeh/metrics-web.git/log/?h=task-24217
15:19:29 <karsten> the level-of-detail question.
15:19:35 <iwakeh> yes
15:19:48 <karsten> I'll check after the meeting. or, more realistically, tomorrow morning.
15:19:54 <iwakeh> I suggested some things at the end of the md doc.
15:19:58 <karsten> thanks for the quick review!
15:20:03 <iwakeh> most importantly,
15:20:24 <iwakeh> I want us to first make the format decision before writing more and
15:20:58 <karsten> reformatting is easy.
15:20:59 <iwakeh> keep in mind that such a doc/spec needs a little math terminology.
15:21:08 <karsten> right.
15:21:13 <iwakeh> which might not be available everywhere.
15:21:19 <iwakeh> formatwise
15:21:25 <karsten> I'm just not sure whether we can come up with one that fits everything before we at least tackled all parts.
15:21:47 <karsten> but really, reformatting is not as time consuming.
15:21:59 <iwakeh> the general format?
15:22:00 <karsten> it's the part where I need to go through really old code.
15:22:23 <iwakeh> maybe, just discuss further on ticket and
15:22:31 <iwakeh> raise the topic again next week here?
15:22:36 <karsten> sounds good.
15:22:56 <karsten> moving on.
15:22:57 <karsten> * Assigned Trac tickets (karsten)
15:23:01 <karsten> https://trac.torproject.org/projects/tor/query?owner=karsten&owner=iwakeh&owner=irl&owner=metrics-team&status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&group=owner&max=300&col=id&col=summary&col=status&col=type&col=priority&col=milestone&col=component&col=changetime&desc=1&order=changetime
15:23:07 <karsten> long link.
15:23:40 <karsten> would you want to go through your tickets after this meeting and unassign those that you're not going to work on over the next week?
15:23:41 <iwakeh> longer list
15:24:19 <irl2> by unassign do you mean have a status not accepted or set owner to metrics-team?
15:24:28 <karsten> set owner to metrics-team.
15:24:47 <iwakeh> ok
15:24:54 <karsten> you could even do that as batch modifification.
15:24:58 <karsten> -fi
15:25:11 <irl2> ok
15:25:20 <karsten> great!
15:25:27 <karsten> * Non-tech/non-slang English translator requirement (karsten)
15:25:42 * iwakeh edited the tilte
15:25:46 <iwakeh> title
15:25:59 <karsten> so, I expect that we're going to write at least 10 times as much text as I wrote for the Servers graphs.
15:26:05 <karsten> probably more.
15:26:12 <iwakeh> surely more.
15:26:39 <karsten> should we ask for somebody to help us with that?
15:26:48 <karsten> or can we handle that on our own?
15:27:11 <karsten> 20 times?
15:27:33 <iwakeh> we could finish an example
15:27:51 <iwakeh> for let's say relays and bridges graphs and find out.
15:28:28 <karsten> the thing is, if we want to ask for resources, we should do it today.
15:28:39 <karsten> because today and tomorrow is when plans are being made.
15:28:44 <karsten> as isabela mentioned last week.
15:28:52 <iwakeh> ah, ok.
15:29:05 <iwakeh> I think there should be support
15:29:18 <iwakeh> for writing things nicely, definitely
15:29:30 <iwakeh> and that person should be involved early to
15:29:38 <karsten> okay. I'll ask for help with that then.
15:29:56 <iwakeh> acquaint themselves with the topics at hand.
15:30:09 <karsten> makes sense.
15:30:25 <karsten> okay, I think that was the last topic on the agenda.
15:30:40 <karsten> anything else? :)
15:30:58 <iwakeh> nope.
15:31:16 <irl2> nothing from me
15:31:39 <karsten> alright! thanks for your time. :) let's talk more via email and next week.
15:31:46 <irl2> ok (:
15:31:49 <karsten> bye, bye!
15:31:52 <irl2> bye!
15:31:56 <iwakeh> bye, bye!
15:31:58 <karsten> #endmeeting