14:59:51 <karsten> #startmeeting metrics team
14:59:51 <MeetBot> Meeting started Thu Jan 12 14:59:51 2017 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:59:51 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:59:55 <karsten> okay, let's start this!
15:00:03 <karsten> https://pad.riseup.net/p/3M7VyrTVgjlF <- agenda pad
15:00:15 <karsten> we already have 8 (!) items, so how about we start right away?
15:00:23 <iwakeh> sure.
15:00:29 <karsten> linda: you around?
15:00:58 <karsten> if not, let's start with the second item.
15:01:23 <karsten> - Atlas/Onionoo usability fixes (karsten)
15:01:40 <karsten> https://trac.torproject.org/projects/tor/wiki/doc/AtlasUsabilityImprovements
15:01:41 <linda> karsten: yep!
15:01:52 <karsten> oh. mind if we do metrics website after this one?
15:02:03 <linda> I don't mind at all. go ahead.
15:02:06 <karsten> ok.
15:02:21 <karsten> that list contains all tickets that are remotely related to the deliverable.
15:02:42 <karsten> since last week we started working on onionoo tickets that would affect atlas usability.
15:02:58 <karsten> including things like being able to search by email address or by ipv6 address without [].
15:03:15 <karsten> those don't require any changes to atlas at all, once they're deployed.
15:03:37 <iwakeh> and improve Onionoo code on the go :-)
15:03:50 <karsten> the plan with atlas tickets is to pick half a dozen or so of them and write patches for those.
15:03:59 <karsten> yes, we did improve onionoo code, too. :)
15:04:22 <karsten> RaBe: that would be the first part of your q1/2017 work.
15:04:33 <linda> yay, RaBe is sticking around!
15:04:38 * linda didn't know this
15:04:42 <RaBe> great! just let me know which tickets you'd prefer :)
15:05:07 <iwakeh> Or, grab as many low hanging (= easy and fast to implement) tickets.
15:05:10 <karsten> linda: yes! we got approval for another contract last week.
15:05:40 <karsten> RaBe: I tried to update many/most of the atlas tickets so that it's relatively obvious what needs to be done.
15:05:57 <karsten> and if I didn't, the reason might be that I didn't find them obvious at all.
15:06:17 <iwakeh> did you tag these tickets?
15:06:29 <karsten> no, I didn't. I just updated them.
15:06:36 <karsten> that list is sorted by modified time.
15:06:44 <iwakeh> Or, even have a milestone on them?
15:06:44 <karsten> would you want to go through them and tag them?
15:06:47 <karsten> nope.
15:07:07 <iwakeh> All Atlas tickets updated recently?
15:07:31 <RaBe> what about the tickets status? could we change the status to "accepted" where i / we should work on? or could i pick any "new" ticket?
15:07:33 <karsten> those seemed doable. except the needs_information ones.
15:07:59 <karsten> "assigned", I think.
15:08:00 <iwakeh> RaBe: do you have a trac account?
15:08:08 <RaBe> not yet
15:08:09 <karsten> "accepted" means that whoever accepts them gets them.
15:08:13 <karsten> (I think.)
15:08:35 <iwakeh> RaBe just register and assign these tickets to yourself.
15:08:43 <RaBe> i could assign tickets i like to myself :D
15:08:53 <karsten> yes, that works. :)
15:09:07 <karsten> want to do that?
15:09:09 <iwakeh> ٩( ᐛ )و
15:09:10 <linda> ᕦ(^o^)ᕤ
15:09:13 <linda> hehehe
15:09:17 <irl> are we on atlas?
15:09:20 <RaBe> i'll register right now while we're talking
15:09:23 <iwakeh> yes
15:09:27 <karsten> oh, hey, irl!
15:10:00 <irl> so i need to apologise and tell you that i'm not going to have time to do any meaningful work on atlas at least for 6 months
15:10:10 <iwakeh> Shall we talk about this list:
15:10:12 <iwakeh> https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&component=Metrics%2FAtlas&col=id&col=summary&col=status&col=owner&col=type&col=priority&col=milestone&col=changetime&desc=1&order=changetime
15:10:31 <irl> any bugs that are currently assigned to me should be considered reassignable
15:10:33 <karsten> irl: what about reviewing and merging? is that something you'd be able to do?
15:10:42 <irl> i can do reviewing and merging/deploying
15:10:58 <irl> very happy to do that
15:11:05 <karsten> should we do that then, so that you'd know what has changed?
15:11:17 <irl> sounds good to me
15:11:35 <iwakeh> great.
15:11:42 <karsten> RaBe: does that work for you? that is, you attach a patch or link to a branch whenever you write code?
15:11:58 <iwakeh> and set it to needs_review
15:12:02 <karsten> yes, and that.
15:12:17 <irl> RaBe: feel free to ping me on irc/email for questions or poking me to review patches
15:12:32 <RaBe> karsten: I think i'll find out how that works :D  (also, i am registered now)
15:12:46 <karsten> perfect!
15:12:55 <linda> :)
15:12:57 <irl> (:
15:13:02 <iwakeh> As RaBe ?
15:13:06 <RaBe> yes
15:13:11 <iwakeh> fine.
15:13:36 <karsten> okay, anything else on the atlas usability improvements topic?
15:13:46 <irl> i could maybe point at a link
15:13:48 <irl> one moment
15:14:01 <irl> https://irl.github.io/atlas/
15:14:10 <irl> https://github.com/irl/atlas
15:14:24 <linda> karsten: hmm. I think I missed the list of stuff.
15:14:25 <irl> RaBe: ^ this was my experiments with usability but they were not really mergable
15:14:38 <irl> sweeping changes that weren't easy to review or test
15:15:04 <RaBe> irl: i'll have a look into that later, thanks!
15:15:05 <irl> RaBe: if you can message me an email address, i'll also dig up a mail thread with feedback from the users
15:15:22 <RaBe> irl: sure
15:15:54 <karsten> iwakeh: I didn't compare your link to the wiki page. how is it different?
15:16:10 * iwakeh didn't compare either
15:16:24 <iwakeh> It's just atlas ticket by most recent changed.
15:16:47 <iwakeh> that is: modified in trac terms.
15:16:56 <karsten> irl: so, the current plan was to work off a few of the trac tickets, rather than do a usability analysis and subsequent improvements.
15:17:03 <karsten> iwakeh: okay, your list sounds good then.
15:17:09 <irl> if it's not a massive pain to do, i'd love a trac report predefined for open atlas issues by priority
15:17:24 <irl> karsten: ok cool, most of the things have trac tickets filed
15:17:42 <karsten> irl: ah, and they are already fixed in your branch?
15:18:04 <karsten> which means that RaBe would have to go find the fixes and turn them into commits to tor's atlas.git master?
15:18:12 <irl> karsten: some things are fixed, other things are "i think it works better this way" and some things are not fixed
15:18:16 <iwakeh> sorting by priority doesn't really reflect our list, does it?
15:18:31 <karsten> no, priorities of current tickets are not very meaningful.
15:18:34 <irl> but potentially some of the fixes just need to be picked out and merged
15:19:06 <irl> i started that with the "how do i find my bridge?" bug, but never finished
15:19:11 <iwakeh> quickly achievable is also a criterium.
15:19:56 <karsten> okay, seems like RaBe should take a look at your branch, in particular before working on tickets that could take more than half an hour or so.
15:19:58 <irl> quickly achievable is a good criteria to start with, perhaps followed by some triage before continuing futher
15:20:07 <irl> karsten: agreed
15:20:50 <karsten> alright. RaBe, I think you have something to start with, and maybe we can talk more next week, together with irl?
15:21:12 <linda> :)
15:21:13 <RaBe> let's to that :)
15:21:16 <irl> sounds good
15:21:18 <karsten> awesome!
15:21:25 <karsten> moving back to the first item:
15:21:29 <karsten> - Metrics website launch (linda)
15:21:32 <karsten> (yay!)
15:21:37 <linda> ٩(^‿^)۶
15:21:57 <linda> There was one person who was pretty insulting in the comments. (T⌓T)
15:21:58 <karsten> thanks for handling all those blog comments, linda!
15:22:09 <linda> No problem. Thanks for helping.
15:22:12 <iwakeh> Reminds me
15:22:21 <iwakeh> how do I get an account there?
15:22:23 <linda> So much spam. And I don't know why some people are so negative.
15:22:31 <karsten> :(
15:22:36 <linda> But besides that guy, everyone else seems fine with it.
15:22:42 <iwakeh> I could only post anonymously.
15:22:47 <karsten> iwakeh: oh, I think I can add users.
15:22:52 <linda> And that guy is objectively wrong.
15:23:00 <linda> iwakeh: yeah, I think someone needs to give you an account.
15:23:01 <iwakeh> troll.
15:23:10 <karsten> iwakeh: but I believe I need to pick your password and you cannot change it. :(
15:23:12 <iwakeh> that person
15:23:29 <linda> iwakeh: should I have just deleted the comment?
15:23:38 <iwakeh> karsten: well, that's fine.
15:23:54 <iwakeh> linda: sure, if it is insulting or the like.
15:23:56 <karsten> user name iwakeh?
15:24:08 <iwakeh> karsten: sounds fine
15:24:41 <karsten> iwakeh: got mail?
15:24:43 <iwakeh> and a 20 character password ;-)
15:25:10 <linda> "I don't know what so-called UX experts are smoking,..."
15:25:13 <iwakeh> fine, looks worse than a 20char pw ;-)
15:25:24 <karsten> same as user name, no?
15:25:34 <iwakeh> well, ad-hominum is always trolling.
15:25:58 <RaBe> has there been any kind of constructive criticism?
15:25:59 <linda> iwakeh: i deleted it and feel better.
15:26:07 <linda> Well, that they didn't like the tabs.
15:26:13 <linda> But lots of calling me stupid in between.
15:26:32 <iwakeh> linda: some newspapers redact and remove such comments stating that people should only post n a polite way.
15:26:35 <linda> I pointed to the document asking for call for design feedback adn told them we didn't liek tabs either and that we know it wont scale
15:26:36 <karsten> other than that, there was some useful feedback.
15:26:57 <karsten> and overall the response was quite positive, I think.
15:26:58 <RaBe> so we ignore the troll :)
15:27:10 <RaBe> great!
15:27:13 <linda> RaBe: yep.
15:27:36 <iwakeh> Just don't feed the troll.
15:27:48 <linda> I never really felt attacked like that before.
15:27:56 <iwakeh> in total, I think it was good feedback.
15:27:58 <karsten> :(
15:28:06 <karsten> sorry, linda. that's the blog.
15:28:09 <linda> I don't keep a personal blog, and even when I post on social media it's my friends.
15:28:10 <iwakeh> linda: I would feel the same.
15:28:29 <linda> and what's more frustrating was that they weren't right.
15:28:37 <iwakeh> linda: these people just feed on making others feel bad.
15:28:44 <linda> I could get over it and take criticism even if handed the wrong way if I can benefit..
15:29:01 <linda> yeah I always heard but never really had to deal with it
15:29:08 <linda> okay. sorry.
15:29:18 <linda> karsten: do you remember any of the feedback that we got?
15:29:29 <linda> karsten: specific ones, I mean, instead of "yay!" "good!" etc.
15:29:47 <karsten> very minor things like "versions should have better colors",
15:29:54 <hiro> linda: aren't you on twitter nowadays? I get insults on a daily basis lately.. some I take as compliments either ;)
15:29:54 <karsten> (which is unrelated to the redesign, but hey)
15:30:07 <hiro> linda: it's good for training
15:30:10 <linda> hiro: I'm not on twitter.
15:30:15 <karsten> new zealand has too few users,
15:30:39 <karsten> but nothing that would require immediate action, in particular with working more on the redesign.
15:30:48 <linda> Oh yeah, making sure that the colors don't repeat is a good thing. But that can be solved with the graphing engine changing.
15:30:49 <karsten> unless I have overlooked something.
15:30:55 <linda> No, that sounds right.
15:31:04 <karsten> linda: there's already a ticket that iwakeh opened.
15:31:06 <linda> And new zealand does have few users.
15:31:13 <linda> karsten: oh, got it.
15:31:55 <karsten> so, it seems to me that we're pretty much done with this project. doesn't mean the site is perfect now, but we're out of time to do more, I think.
15:32:03 <linda> karsten: agreed.
15:32:04 <iwakeh> linda: I witnessed that many times in mailing lists. sigh.
15:32:24 <linda> iwakeh: (>﹏<)
15:32:59 <karsten> I'll also look out for blog comments.
15:33:14 <linda> karsten: thanks!
15:33:35 <linda> how long should the comments be open for?
15:33:37 <karsten> though I think there won't be as many from now on, because there are newer blog posts now.
15:33:49 <karsten> 1 week total?
15:34:06 <linda> karsten: sounds good to me.
15:34:38 <karsten> alright!
15:34:43 <karsten> moving on?
15:34:59 <karsten> ah, before we move on:
15:35:16 <karsten> regardless of negative people on the blog, I found this redesign project very successful!
15:35:32 <karsten> I'm really happy how it went. thanks to everyone involved!
15:35:44 <iwakeh_> I lost the connection ...
15:35:45 <karsten> not giving blog commenters the last word. hah.
15:36:08 <karsten> iwakeh: we only discussed that blog comments should be open for 1 week in total.
15:36:17 <iwakeh_> fine.
15:36:23 <karsten> let's move on, still so much on the agenda. :)
15:36:25 <RaBe> i really liked working with this great team (and i'm happy we continue our work!) :D
15:36:32 <karsten> yay, RaBe! :)
15:36:38 <karsten> speaking of....
15:36:39 <karsten> - Web design plans for Q1/2017 (karsten)
15:36:41 <karsten> :)
15:36:52 <linda> ^_^
15:37:03 <linda> RaBe: you were awesome
15:37:15 <karsten> so, that plan had 5 pieces.
15:37:18 <linda> karsten: you came up with so many ideas, inclucing the layout, second header, etc.
15:37:25 <linda> iwakeh: always the voice of reason and practical
15:37:27 <linda> <3 <3
15:37:37 <linda> okay i'll stop you have other things to talk about hehe
15:37:39 <karsten> yes, pretty good team indeed!
15:38:09 <karsten> so, part 1 of the plan was atlas usability improvements. we covered that.
15:38:14 * iwakeh_ struggling to reclaim 'iwakeh'
15:38:38 <iwakeh_> The time limit, too?
15:38:42 <karsten> part 2 is metrics website improvements. it seems there won't be much left, but who knows what comments we get this weekend...
15:39:30 <karsten> ah, the time limit. well, we discussed whether there should be a limit of, say, 8 hours of RaBe's time for atlas improvements.
15:39:57 <iwakeh_> otherwise this becomes endless.
15:40:11 <karsten> right.
15:40:23 <karsten> okay, let's go through the other parts and then think about time.
15:40:40 <karsten> part 3 is a CSS template for our javadocs.
15:40:45 <iwakeh_> karsten: Could you paste them on the pad?
15:41:03 <karsten> ah!
15:41:03 <RaBe> i'm not familiar with javadocs, but i do know css :)
15:41:13 <iwakeh_> that should suffice.
15:41:16 <karsten> iwakeh_: can you, while I find javadocs as example?
15:41:24 * iwakeh_ searching ..
15:41:49 <irl> http://irl.sdf.org/strigidoc/apidocs/
15:41:51 <irl> example javadoc
15:42:06 <iwakeh> all fine, that needs to be talked about.
15:42:17 <karsten> irl: thanks. :)
15:42:24 <RaBe> so you just want some css to make it a bit less ugly?  :D
15:42:33 <linda> RaBe: lol :)
15:42:39 <iwakeh_> Right :-)
15:42:48 <iwakeh_> and confirm to Metrics style.
15:42:58 <iwakeh_> as far as reasonable.
15:43:01 <RaBe> is it one css for all your javadocs, or do you need individual css filesß
15:43:02 <RaBe> ?
15:43:03 <linda> I would love it if it looked more consistent with other tor stuff.
15:43:26 <iwakeh_> RaBe: yes
15:43:30 <iwakeh_> more here
15:43:32 <iwakeh_> http://www.oracle.com/technetwork/articles/java/index-jsp-135444.html
15:43:35 <linda> But the style guide has yet to be applied to anything other than metrics.. and we don't have a central place for all our media, and torproject.org badly needs work, and...
15:43:48 <linda> so just yeah, make it look better and make it look like it is obviously associated with metrics.
15:43:56 <karsten> https://people.torproject.org/~karsten/volatile/javadoc/
15:43:58 <iwakeh_> linda: +1
15:44:12 <RaBe> so i'd need a list with all the javadocs you'd like to update
15:44:21 <karsten> ^ that are our javadocs for metrics-lib.
15:45:04 <iwakeh_> it's our three releases.
15:45:35 <iwakeh_> https://dist.torproject.org/
15:45:44 <iwakeh_> collector, descriptor, onionoo
15:45:55 <iwakeh_> should be the same css for all.
15:45:57 <karsten> RaBe: okay, I'd say take a look at these resources, and let's talk more next week.
15:46:04 <iwakeh_> yes.
15:46:05 <RaBe> okay!
15:46:21 <karsten> let me quickly finish the 5 parts.
15:46:32 <karsten> 4 is a project website for metrics-lib, which is probably small.
15:46:44 <karsten> 5 is adapting collector and onionoo to the metrics website style.
15:47:07 <karsten> I sent this plan in german to RaBe the other day.
15:48:07 <karsten> I think what we need to decide is whether this plan sounds sane,
15:48:17 <karsten> and how you could split your hours on the various parts.
15:48:44 <karsten> how about we take that to next week's meeting?
15:48:50 <iwakeh_> yep.
15:49:00 * linda is excited for this work
15:49:22 <karsten> yes, this will be pretty cool!
15:49:25 <iwakeh_> linda: could be the predecessor for other code docs at tor.
15:50:04 <karsten> we should move on.
15:50:10 <karsten> so many topics..
15:50:19 <iwakeh_> next :-)
15:50:23 <karsten> - OnionPerf review and deployment (hiro)
15:50:44 <hiro> yes :) I sent what I have been doing up to now to Rob just today
15:51:00 <karsten> saw that. great!
15:51:01 <hiro> I have been documenting onionperf, installation process and data
15:51:06 <linda> iwakeh_: +1
15:51:24 <hiro> now the next step for me would be writing a short summary of what information we have from each metric
15:51:36 <hiro> and collecting feedback on the work so far
15:51:42 <iwakeh_> cool.
15:52:02 <hiro> here is the readme up to now: https://github.com/robgjansen/onionperf/compare/master...hiromipaw:docs?expand=1
15:52:54 <karsten> one thing to figure out (now or later) is where to run these onionperf instances.
15:53:06 * irl needs to disappear for another meeting now, will read backlog for any highlights if i'm implicated in anything, otherwise see you next week
15:53:14 <karsten> sounds good. thanks, irl!
15:53:14 <hiro> see you irl :)
15:53:17 <iwakeh_> irl: bye. bye.
15:53:41 <karsten> hiro: I assume one instance will run on a hetzner (virtual) host,
15:53:49 <karsten> but do you have any plans for running more instances?
15:54:00 <hiro> karsten: at the moment I have been running this locally
15:54:17 <karsten> sure. that should work fine for testing.
15:54:37 <hiro> yep, if we have a machine that we can use, I'll be happy to install it
15:54:39 <karsten> I'm just not sure how long it will take to find other places to host this.
15:55:02 <iwakeh_> It won't need much bandwidth?
15:55:07 <karsten> should I talk to sysadmins for this, or do you want to do that?
15:55:25 <hiro> karsten, I can open a ticket and put you on cc so you can follow
15:55:49 <karsten> hiro: sounds good, please do.
15:55:57 <hiro> sure will do as we end this meeting :)
15:55:59 <karsten> iwakeh_: not much bandwidth, no.
15:56:03 <karsten> great!
15:56:03 <iwakeh_> How many shall we have?
15:56:11 <karsten> so far we had 3 torperf instances.
15:56:30 <iwakeh_> Where are these running?
15:56:34 <karsten> but ideally distributed a bit.
15:57:06 <karsten> one on a tor vm, one on moria at mit, one by linus in, err, scandinavia.
15:57:32 <hiro> karsten: regarding this machine, how often can we delete logs? Do we need to keep historical data?
15:57:42 <iwakeh_> so, currently asia, africa, australia missing?
15:58:29 <karsten> hiro: it wouldn't hurt to keep logs for a bit. after all, it's our data that we're generating here, not the data of users.
15:58:57 <hiro> yep but I was thinking only in terms of space... like 1 month would it be ok?
15:59:02 <karsten> iwakeh_: not sure how diverse we should get here. I was thinking to find several well-connected places to have some more robustness.
15:59:21 <karsten> hiro: or several months. or does onionperf log excessively?
15:59:43 <karsten> isn't disk space cheap?..
16:00:10 <hiro> no it doesn't that much...
16:00:34 <hiro> there are a number of lines every our... but not many
16:00:41 <iwakeh_> karsten: Would be interesting to know the differences all over the world, but could be a future extension.
16:00:52 <karsten> iwakeh_: true!
16:01:08 <karsten> the first step would be to replace our torperfs.
16:01:13 <iwakeh_> right.
16:01:20 <karsten> though not necessarily on the exact same machines.
16:01:38 <karsten> okay, we just crossed the 60 minute line.
16:01:43 <karsten> with four topics left.
16:01:53 <hiro> iwakeh_: maybe we could use something like ec2 and distribute this in different continents? would that be expensive? (I honestly do not really know yet)
16:01:58 <karsten> can we move them to next week?
16:02:09 <iwakeh_> sure.
16:02:12 <hiro> fine by me
16:02:25 <karsten> hiro: ec2 might be an option.
16:02:59 <karsten> I don't know how expensive this would be. would you want to try out setting it up on a tiny instance there?
16:02:59 <iwakeh_> yes, but that might skew data maybe?
16:03:07 <karsten> why?
16:03:23 <iwakeh_> Well, traffic might be reduced in these centers?
16:03:40 <iwakeh_> I didn't use these yet, so just guessing.
16:03:58 <iwakeh_> It's would seem more realistic to use a regular machine
16:04:11 <hiro> I thought we might get better performances instead
16:04:12 <iwakeh_> outside a known ec2 center ip-range.
16:04:40 <karsten> I think performance should be fine. without knowing, just a guess.
16:04:55 <karsten> the advantage of ec2 is indeed that it's easy to distribute instances.
16:04:57 <iwakeh_> We intend to measure client experience?
16:05:15 <karsten> and hey, we have hiro with her magic docker-or-ansible powers who could automate this like hell!
16:05:33 <karsten> well, client experience...
16:05:38 <iwakeh_> hey, have movable OnionPerfs.
16:05:38 <hiro> haha yeah I can make containers
16:05:55 <karsten> it's more that we're trying to measure performance consistently to detect trend changes.
16:06:08 <karsten> the current torperfs are run on well-connected hosts. those are unlike clients, I guess.
16:06:23 <iwakeh_> It#S an important question.
16:06:37 <iwakeh_> for just detecting changes ec2 is fine.
16:07:10 <karsten> yes, this seems like a good question for the next step after replacing torperf.
16:07:26 <karsten> how about we compare prices of ec2 vs. rented vhosts?
16:07:40 <karsten> with the additional complexity of renting vhosts in different places taken into account.
16:07:54 <karsten> I mean, tor already has an ec2 account.
16:08:15 <karsten> and maybe this runs on a tiny/micro/nano/whatever-they-call-it instance and is really cheap.
16:08:21 <hiro> I can start doing that
16:08:27 <karsten> that would be neat!
16:08:33 <iwakeh_> yes!
16:09:01 <karsten> and the automation part was not a joke. that would be pretty useful.
16:09:15 <iwakeh_> definitely!
16:09:34 <iwakeh_> reproducibility, too.
16:09:36 <karsten> okay, 70 minutes into the meeting.
16:09:38 <karsten> yes!
16:09:42 <hiro> yeah that shouldn't be too complicated
16:09:54 <karsten> depending on who you ask. ;)
16:10:19 <karsten> alright, let's move the other topics to next week. all good topics, but we're out of time.
16:10:30 <karsten> to next week or email.
16:10:40 <iwakeh_> let' just work on these topics :-)
16:10:55 <karsten> end of meeting, or is there anything that requires discussion now?
16:11:03 <iwakeh_> nope, all set.
16:11:04 <hiro> personally I am good.
16:11:24 <karsten> cool! thank you, and bye!
16:11:30 <karsten> #endmeeting