14:59:46 <karsten> #startmeeting metrics team
14:59:46 <MeetBot> Meeting started Thu Jan 26 14:59:46 2017 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:59:46 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:59:55 <karsten> https://pad.riseup.net/p/3M7VyrTVgjlF <- agenda pad
15:00:07 <iwakeh> hi there!
15:00:12 <karsten> hi iwakeh!
15:00:28 <hiro> hi iwakeh could you pass me a link to your gpg key?
15:00:30 <karsten> the pad already contains a few topics and details.
15:00:36 <hiro> apparently the one I used yesterday is expired
15:00:39 <iwakeh> sure, looking...
15:00:55 <iwakeh> hiro: I'll just mail it to you later? ok?
15:01:03 <hiro> sure :)
15:02:05 <karsten> shall we just start?
15:02:11 <karsten> we can always append new topics to the agenda.
15:02:21 <RaBe> go go go :)
15:02:25 <karsten> going!
15:02:29 <karsten> - Web design (RaBe)
15:02:44 <karsten> I already put a few subtopics on the pad.
15:02:52 <karsten> want to go through them?
15:03:03 <karsten> or shall I?
15:03:15 <RaBe> i assigned the atlas tickets i'm planning to work on to myself. unfortunately i had another important project this week, so i did not work on them...
15:03:31 <karsten> (no worries, this is not super urgent)
15:03:57 <karsten> do you have any questions about them that are easier to discuss here than on the ticket?
15:04:22 <RaBe> not at the moment :) when i read them the first time, i think everything was clear
15:04:28 <karsten> cool!
15:05:58 <karsten> I put some questions on the pad, which come from an earlier update from irl.
15:06:06 <karsten> let me just repeat them here.
15:06:11 <karsten> Can somebody who is not irl review #21291?
15:06:42 <RaBe> i can review #21291 if you want, since i already had been working on the flags... also i'll have to merge it to my tooltip ticket
15:06:57 <karsten> I think that would be great!
15:07:14 <iwakeh> good.
15:07:22 <karsten> Should we not implement #19452, because currently we have high cache hit rates on onionoo and this would result in more complex, less cacheable requests?  If so, should we close that ticket as wontfix with this explanation, together with a copy of https://volatile.noreply.org/2017-01-24-2055rzmtvB0/varnish_cache-year.png ?
15:07:37 <karsten> I think the reasoning is plausible.
15:08:01 <RaBe> i'm not sure about that one, but each closed ticket is great :D
15:08:05 <karsten> haha
15:08:14 <karsten> I'm not sure, either.
15:08:23 <karsten> but we'd probably have to run some performance tests to be sure.
15:08:37 <karsten> and maybe we don't have to, unless this some day shows up as bottleneck.
15:08:59 <RaBe> that ticket seems like a lot of work...
15:09:34 <karsten> so, should we not fix it then?
15:09:56 <karsten> something I wonder is whether we'll find the closed ticket if this idea ever resurfaces.
15:09:56 <RaBe> i can't answer the performance question. if it IS an issue, we should fix it anyway.
15:10:15 <karsten> well, the argument is that our fix might perform worse.
15:10:15 <iwakeh> There are those trac suggestions ;-)
15:10:31 <karsten> heh, did you ever receive a useful trac suggestion?
15:10:49 <iwakeh> I always have new ticket topics :-)
15:10:53 <karsten> hah
15:10:57 <karsten> okay, close?
15:11:10 <RaBe> i won't stop you :)
15:11:23 <karsten> will do.
15:11:27 <iwakeh> if it comes up agian we can deal with it.
15:11:30 <hiro> :)
15:11:32 <karsten> great!
15:11:41 <karsten> Are there objections to removing graphs with no data as discussed in #19553?  If not, irl will prepare a patch tonight.
15:12:01 <hiro> not on my side
15:12:08 <hiro> I think it actually makes sense
15:12:29 <RaBe> yes (what means, no objections)
15:12:32 <karsten> ok. and RaBe, would you review that patch afterwards?
15:12:45 <RaBe> sure
15:12:54 <karsten> thanks!
15:13:02 <karsten> okay, I think that's all for atlas.
15:13:17 <karsten> hmm, maybe we should discuss onionoo next.
15:13:24 <iwakeh> fine.
15:13:38 <karsten> - Onionoo release (karsten)
15:13:46 <karsten> how about we put out a release tomorrow or saturday?
15:14:10 <karsten> with the reason being that we should do this in january, and next week might be hectic.
15:14:15 <karsten> (berlin meeting)
15:14:19 <karsten> (possibly)
15:14:23 <iwakeh> https://trac.torproject.org/projects/tor/milestone/Onionoo%203.2-1.1.0
15:14:45 <karsten> right. I have some suggestions on that list.
15:14:55 <karsten> - keep #16552 scheduled for 4.0-?.?.?
15:14:55 <karsten> - review #17938 for 3.2-1.1.0
15:14:55 <karsten> - add #20994 to 3.2-1.1.0
15:14:56 <karsten> - remove #21138 from 3.2-1.1.0
15:17:08 <karsten> (there's activity on the pad :))
15:17:28 <iwakeh> (strange)
15:18:46 <karsten> okay, sounds like we agree on the tickets.
15:18:50 <karsten> how about the timing?
15:19:03 <iwakeh> aim at friday?
15:19:11 <karsten> works for me.
15:19:33 <karsten> cool!
15:19:41 <karsten> sounds like we have a plan for that.
15:19:43 <karsten> next topic?
15:19:47 <iwakeh> I'm wondering about naming the onionoo milestones without protocol?
15:20:01 <karsten> hmmm
15:20:11 <karsten> yeah, that might be better.
15:20:11 <iwakeh> To avoid the regrouping.  The release number ought to stay the same.
15:20:27 <iwakeh> ok.
15:20:34 <karsten> want to make that change?
15:20:57 <iwakeh> ok.
15:21:06 <karsten> ah, so #16552,
15:21:14 <karsten> that's a rather small change,
15:21:31 <karsten> but it requires a new major protocol version number.
15:21:40 <karsten> because of backward-incompatibility.
15:21:40 <iwakeh> depends ...
15:22:03 <karsten> so, the idea was to announce that we'll release it in 1 month.
15:22:18 <iwakeh> Onionoo-1.1.1
15:22:31 <iwakeh> release 4.0-1.1.1
15:22:42 <karsten> hmmmm
15:22:53 <karsten> or maybe 1.2.0 was okay, too. I'm confused. :)
15:22:56 <iwakeh> release 4.0-1.2.0
15:23:14 <karsten> yes, I think that's good.
15:23:17 <karsten> sorry :)
15:23:18 <iwakeh> Better to be moredramatic
15:23:40 <karsten> alright, moving on to less confusing topics?
15:23:51 <karsten> (hopefully)
15:23:53 <iwakeh> ok
15:23:55 <karsten> - More web design (RaBe)
15:24:15 <karsten> RaBe: I put in "Status of Javadoc CSS" here, but is there anything else that you need input on?
15:24:28 <karsten> if not, that's fine.
15:24:37 <karsten> but maybe now's a good time.
15:24:42 <RaBe> i could not work on the javadocs either, i'm sorry... has been a busy week :) but i know what to do until the next meeting!
15:24:49 <karsten> perfect!
15:25:00 <karsten> moving on:
15:25:03 <karsten> - OnionPerf (hiro)
15:25:13 <karsten> I can see a directory listing with files. :)
15:25:43 <hiro> yep
15:25:52 <karsten> hiro: did you compare these results with the torperf results we have?
15:25:56 <hiro> sp that's the deployment on otf cloud
15:26:22 <karsten> cool!
15:26:48 <hiro> the machine on tpo infrastructure is ready to, I was planning to finish installing it this week
15:26:48 <karsten> how hard was it to set this up? and how hard would it be to set up more in the future?
15:26:55 <hiro> not very hard
15:27:08 <karsten> great!
15:27:16 <hiro> so I cannot use docker on OTF cloud for example, but I haven't tried ansible
15:27:33 <hiro> in any case I could easily do a small bash script that automates everything
15:27:41 <hiro> there aren't too many packages to install
15:27:57 <karsten> ok.
15:27:58 <hiro> would you like me to document this somewhere?
15:28:10 <karsten> might turn out to be useful, yes.
15:28:15 <hiro> ok perfect
15:28:28 <hiro> regarding your question, I haven't compared results just yet
15:28:44 <karsten> would you want to try that?
15:28:54 <karsten> or should somebody else make a start there?
15:29:06 <iwakeh> or, next week?
15:29:20 <iwakeh> during the meeting?
15:29:34 <karsten> if that happens, yes, why not!
15:29:38 <hiro> so what I can tell you is that the tpf files are generated according to the torperf specs
15:29:44 <hiro> I mean they have the same fields inside
15:29:54 <hiro> I haven't checked the json though
15:30:08 <karsten> we might not need the json.
15:30:13 <hiro> and yes I can start to do that
15:30:21 <karsten> well, up to you.
15:30:29 <hiro> regarding next week meeting that's a bit too soon for me
15:30:35 <karsten> okay.
15:30:37 <hiro> feb 13 might work better
15:31:18 * iwakeh looking at my calendar
15:32:10 <karsten> hiro: do you know how to start?
15:32:17 <iwakeh> what about fri 10.2.?
15:32:47 <karsten> maybe. I'll have to look at my calendar after this meeting.
15:32:50 <hiro> Iwakeh: might work too I have to check tickets for price and such
15:33:14 <karsten> hiro: like, if you're using java, you can use metrics-lib to parse .tpf files.
15:33:29 <karsten> hiro: otherwise you'll have to write your own parser, which is certainly doable.
15:33:36 <hiro> so karsten my idea would be to start parsing the files and maybe plotting them and see
15:33:53 <hiro> I think I could use the python library too
15:34:05 <karsten> stem? does that support .tpf files?
15:34:14 <iwakeh> no idea?
15:34:16 <hiro> not sure, I will look it upo
15:34:28 <karsten> https://stem.torproject.org/api.html#descriptors
15:34:33 <karsten> doesn't look like it.
15:34:46 <karsten> but again, writing your own parser for this analysis is not a crazy amount of work.
15:35:07 <karsten> anyway, please shout if you need help with this!
15:35:25 <hiro> sure
15:35:39 <hiro> and also if I write it I could try to add it to stem
15:35:42 <hiro> let's see
15:36:13 <karsten> hmmmm, I'd say don't bother, as we might not keep that format forever.
15:36:34 <karsten> whatever gets the job done.
15:36:43 <hiro> ok
15:36:51 <iwakeh> maybe jython and metrics-lib?
15:37:23 <iwakeh> http://www.jython.org/
15:37:37 <iwakeh> for a quick and easy approach?
15:37:59 <hiro> I'll try, thanks iwakeh
15:38:04 <iwakeh> but, you see what works best for you.
15:38:04 <karsten> cool!
15:38:08 <karsten> moving on?
15:38:20 <iwakeh> ok
15:38:39 <karsten> - Tech reports (iwakeh)
15:38:50 <iwakeh> in the making.
15:38:58 <karsten> I'll prepare a db dump as discussed on the pad later today.
15:39:08 <iwakeh> that'll be great!
15:39:20 <karsten> but that will take a bit. query without useful index, compress, download, upload...
15:39:47 <iwakeh> is it too laborious
15:39:53 <iwakeh> ?
15:40:00 <karsten> for me? no.
15:40:04 <karsten> it just keeps machines busy.
15:40:14 <karsten> no worries, I'll do this.
15:40:17 <iwakeh> well, that's what they are made for.
15:40:24 <iwakeh> thanks!
15:40:27 <karsten> hehe. sure!
15:40:52 <karsten> moving on?
15:41:05 <iwakeh> yep.
15:41:11 <karsten> - Tor Browser downloads and updates (karsten)
15:41:18 <karsten> https://metrics.torproject.org/webstats-tb.html
15:41:27 <karsten> not linked from the homepage yet.
15:41:43 <karsten> discussion is/was on #21236.
15:41:48 <iwakeh> Is that the reason for the missing part after 'Home>>'?
15:41:54 <karsten> yes.
15:41:59 <karsten> (I hope so!)
15:42:00 <iwakeh> ok.
15:42:03 <karsten> I was wondering, too.
15:42:28 <karsten> btw, which logs did you parse?
15:42:45 <iwakeh> whatever could be reached.
15:42:46 <karsten> I wonder if I should just re-parse them to find that log line that matches with fewer groups.
15:43:07 <RaBe> there's just an empty <li></li> where the title of the graph is missing :)
15:43:39 <karsten> line?
15:43:49 <RaBe> 154
15:44:02 <RaBe> <li class="active">
15:44:31 <RaBe> also, the next container is missing its headline and description (line 158 ff )
15:44:32 <karsten> right!
15:44:36 <karsten> and that.
15:44:38 <iwakeh> oh right!
15:45:01 <karsten> I'd have to look at the code to say whether that will be filled in.
15:45:06 <karsten> but my current plan is to just try it out. ;)
15:45:23 <karsten> but, not before we're sure we want to publish this graph.
15:45:24 <iwakeh> karsten: I think it was one of the humongous lines ...
15:46:05 <karsten> yeah, could be. :)
15:46:16 <karsten> ah! you mean the log.
15:46:23 <iwakeh> yes.
15:46:34 <karsten> thought you meant the jsp.
15:46:36 <iwakeh> but why is the title of the graph missing?
15:47:04 <karsten> because it's not listed in categories.json.
15:47:31 <karsten> in the deployed version.
15:47:55 <karsten> no need to go bug hunting yet. it's just a hack to get this graph published without linking it from the home page.
15:48:22 <karsten> so,
15:48:27 <karsten> - tor-project@ post Fri or Sat or prepare blog post for Mon or Tue?  can also ask at Vegas meeting
15:48:34 <karsten> what do you think?
15:48:55 <iwakeh> why not?
15:49:15 <karsten> I guess press people would say fri/sat gets almost no attention, whereas mon/tue gets a lot.
15:49:27 <iwakeh> ah
15:49:47 <karsten> and if we want to do a blog post, we'll have to write it.
15:49:51 <iwakeh> maybe get suggestions from the vegas meeting?
15:49:56 <karsten> ok.
15:50:13 <iwakeh> gk said it looks good.
15:50:14 <karsten> in any case, I'd want to get this done by tue, so that it's still in january.
15:50:19 <karsten> yep.
15:50:43 <iwakeh> fine.
15:50:53 <karsten> but he only saw 4 data points when he said that, not over a year. ;)
15:51:10 <karsten> moving on?
15:51:11 <iwakeh> well, it's about the graph not the data ;-)
15:51:14 <karsten> heh
15:51:39 <karsten> though I guess we already talked about this:
15:51:40 <karsten> - Berlin meeting (karsten)
15:51:53 <karsten> I'll check feb 10 in a few minutes.
15:52:10 <karsten> would that work for you, iwakeh and hiro?
15:52:13 <iwakeh> actually its berlinale.
15:52:32 <iwakeh> that might be problematic with travelling?
15:53:10 <karsten> we should check.
15:53:25 <hiro> ok I'll check it
15:53:34 <karsten> ok. more via email?
15:53:48 <iwakeh> https://www.berlinale.de/en/globals/festivaldaten/Eckdaten.html
15:53:55 <iwakeh> 9.2-19.2.
15:54:07 <iwakeh> ok.
15:54:33 <karsten> out of topics!
15:54:45 <iwakeh> and out of time :-)
15:54:50 <karsten> does that mean we can finish 5 minutes early?
15:54:54 <RaBe> :D
15:54:55 <karsten> almost out of time.
15:55:08 <iwakeh> oh, space time dilation ... fine :-)
15:55:27 <hiro> hehe
15:55:33 <karsten> yay. thanks, everyone! ending the meeting...
15:55:35 <hiro> I'll check the tickets and get back to you
15:55:36 <karsten> #endmeeting