16:01:17 <karsten> #startmeeting metrics team
16:01:17 <MeetBot> Meeting started Thu Mar 28 16:01:17 2019 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:17 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:01:31 <karsten> - TorDNSEL (karsten)
16:01:36 <karsten> haskell.
16:01:47 <karsten> so, I made progress.
16:01:58 <irl> that is better than expected (:
16:01:59 <karsten> but I can't really say how much is left to do.
16:02:13 <karsten> I had help. :)
16:02:31 <irl> so i said to ln5 earlier today, that i'm not sure which of the two solutions would be ready first
16:02:39 <karsten> yeah.
16:03:02 <karsten> and both have their own risks.
16:03:18 <karsten> these haskell fixes might turn out to compile and work okay in most cases but not all.
16:03:40 <karsten> the rewrite might work fine for the cases we thought about, but not for the cases that tup thought about long ago and that we forgot.
16:03:41 <irl> right
16:04:19 <karsten> so, I guess that neither of the two solutions will be ready by end of this week.
16:04:46 <irl> i don't think i'll be making any progress this week, i have lots of admin stuff to catch up on
16:04:57 <karsten> makes sense.
16:05:16 <karsten> I think I'd want to continue with the haskell upgrade project until I feel I'm not making any progress anymore.
16:05:33 <irl> it would be great to run both systems in paralell for a while if we can
16:05:39 <karsten> true!
16:06:01 <karsten> how about we ask ln5 for mid-april as new deadline?
16:06:22 <irl> how many points is that?
16:06:34 <karsten> hmmmmmm
16:07:08 <karsten> it's been a while since I last looked at the roadmap. (when is gaba back?...)
16:07:09 <irl> ideally we would have 8 points, but could probably live with 5
16:07:47 <irl> i don't know how many you think you might need for the haskell work though
16:08:01 <irl> and we lose the ability to run both systems together if we don't do the haskell work before the cutoff
16:08:49 <karsten> this is hard to express in points.
16:09:10 <karsten> in theory, we could define an upper limit and stop the work, regardless of success or not.
16:09:40 <karsten> but estimating how much work there is to do is impossible for me.
16:09:50 <karsten> I mean, I don't understand most of what the code does.
16:10:36 <irl> hm ok
16:10:52 <karsten> so,
16:11:01 <irl> my points estimates are based on what i estimated in the roadmap but may not actually be accurate
16:11:18 <karsten> how about we keep working on this for another week, in parallel.
16:11:32 <irl> sounds good, then we can see how things look
16:11:45 <karsten> yes.
16:11:53 <karsten> we have a few alternatives, though.
16:12:06 <karsten> we could run the current code on squeeze if we want to have a second instance.
16:12:11 <karsten> on our own squeeze, that is.
16:12:33 <karsten> and if everything falls apart, we might survive without a running instance, even if that'd be sad.
16:13:11 <karsten> but let's see how things evolve until next thursday.
16:13:17 <irl> ok
16:13:22 <karsten> it's great that the sponsor 13 work is done, by the way.
16:13:26 <karsten> that was really important!
16:13:40 <irl> yes (:
16:13:43 <karsten> so, I think it's okay that we didn't complete this tordnsel thing on schedule.
16:14:17 <karsten> ah, regarding the haskell upgrade: do we know anybody in tor who writes haskell?
16:14:34 <karsten> or wrote? or has been less active lately? somebody who I could ping?
16:15:04 <irl> i don't know of anyone
16:15:29 <karsten> okay.
16:15:49 <karsten> alright, moving on?
16:15:58 <irl> ok
16:16:02 <karsten> - Performance measurements/analysis (karsten)
16:16:12 <karsten> I think we'll have another meeting next week.
16:16:32 <karsten> or at least we said in the meeting three weeks ago that we'll have another one four weeks later.
16:17:11 <karsten> would you be able to upgrade the tor versions on the op-* instances until then?
16:17:34 <karsten> mostly to show that we're making progress.
16:17:40 <karsten> along with the tickets I created.
16:17:50 <irl> possibly
16:18:28 <karsten> ok.
16:18:49 <irl> is this to move them to a stable version?
16:18:56 <irl> do we know what version we want to test?
16:19:03 <karsten> something recent.
16:19:08 <irl> i guess that the current "whatever is head of master" is wrong
16:19:10 <karsten> maybe what's shipped in debian?
16:19:19 <karsten> yes, that's very recent.
16:19:42 <karsten> you're building from source?
16:20:22 <irl> it is built from source currently
16:20:30 <irl> but i want to change it to use system tor instead from deb pkg
16:20:45 <irl> the decision is whether we use the long-term stable in debian archives, or latest stable in deb.tpo
16:20:48 <karsten> I don't feel strongly which exact version. but it would be good not to have to argue whether an issue also exists in recent tor versions or not.
16:21:10 <irl> let's do the deb.tpo one then
16:21:11 <karsten> whichever is easier.
16:21:16 <karsten> IMO.
16:21:26 <irl> basically the same
16:21:31 <karsten> ok.
16:21:51 <karsten> sounds good.
16:22:02 <karsten> moving on?
16:22:05 <irl> acute: did you want to say anything on onionperf?
16:22:12 <karsten> ah.
16:22:19 <acute> hi
16:22:24 <karsten> hi!
16:23:04 <acute> just that we can currently close some more bugs
16:23:22 <karsten> sounds great! :)
16:23:35 <acute> also, I've updated the deployment instructions in case you want to also update the op instances themselves
16:23:56 <acute> this will allow you to measure v3 services also
16:24:19 <karsten> neat!
16:24:23 <karsten> how does the scheduling work there?
16:24:33 <karsten> does it round-robin between public, v2, and v3?
16:24:40 <acute> yes, exactly
16:24:45 <irl> we might need to make changes to how we do stats for the output for that one
16:24:55 <irl> i guess we have some regex of .onion and now we need to look at length
16:25:12 <karsten> right. but that's easy. still good to keep in mind. ;)
16:25:23 <karsten> any reason not to add v3 now?
16:25:36 <karsten> or maybe on one instance?
16:26:09 <irl> acute: are you running a test instance?
16:26:10 <karsten> do you happen to have a .tpf file with v2 and v3 measurements?
16:26:26 <acute> yes, but currently only measuring bridges
16:26:46 <acute> ah - also for v3 you need a backported version of stem (1.7)
16:26:54 <acute> yes, can provide this
16:27:04 <karsten> just for the regex.
16:27:16 <karsten> can you email me that, or put it on a ticket somewhere?
16:27:25 <acute> yep will email you
16:27:30 <karsten> great!
16:27:38 <acute> :)
16:28:32 <karsten> what do you think about deploying on one instance?
16:28:52 <karsten> maybe next week or so.
16:29:14 <irl> acute: do you think it's stable enough to set it up on op-ab?
16:29:24 <acute> yes, I do
16:29:34 <irl> ok let's do that
16:29:48 <karsten> isn't there an end-of-month thing regarding funding?
16:30:02 <karsten> would it make sense to deploy before end of month in order to include it in a report somewhere?
16:30:20 <irl> excellent point
16:31:25 <karsten> okay, I can make sure that metrics-web doesn't choke on the v3 measurements tonight.
16:31:39 <karsten> the .tpf file will help with that.
16:32:16 <karsten> cool! moving on to the next topic?
16:32:22 <irl> ok
16:32:29 <karsten> - Next meeting time (karsten)
16:32:50 <karsten> next week clocks will change again.
16:33:23 <irl> maybe for the last time
16:33:33 <karsten> I could do this meeting at the same UTC time. but it might collide with another meeting then.
16:33:42 <karsten> unclear if for the last time. :(
16:33:52 <karsten> but what's certain in politics...
16:34:20 <karsten> I'll learn more about the other meeting's time today. how about we talk about maybe moving this meeting time via email?
16:34:32 <irl> that sounds good, i'm thinking i don't know about other meetings
16:35:05 <karsten> okay.
16:35:19 <karsten> the next two topics are gaba topics.
16:35:22 <karsten> or at least require gaba.
16:35:57 <karsten> we could try signal and ask her.
16:36:03 <irl> sounds good
16:36:29 <karsten> alright!
16:36:34 <karsten> out of topics?
16:36:55 <irl> no more from me
16:37:09 <karsten> great!
16:37:30 <karsten> thanks, and have a good evening! bye!
16:37:32 <irl> bye!
16:37:41 <karsten> #endmeeting