15:00:08 <karsten> #startmeeting metrics team
15:00:08 <MeetBot> Meeting started Thu May  2 15:00:08 2019 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:08 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:00:15 <gaba> hi!
15:00:36 <karsten> hi gaba!
15:00:57 <gaba> so funny I got confused :)
15:01:04 <gaba> good that it is 'just' weekly meeting
15:01:09 <karsten> haha
15:01:24 <irl> gaba: it's ok, i had the jitsi set up and ready to go too
15:01:39 <gaba> :)
15:02:03 * karsten tries to remember what else needs discussion...
15:03:05 <karsten> anyway, agenda is complete?
15:03:27 <acute> nothing else from me
15:04:06 <karsten> okay, let's start then.
15:04:11 <karsten> - Research portal. What are we missing for new one? https://trac.torproject.org/projects/tor/ticket/26838
15:04:25 <irl> https://trac.torproject.org/projects/tor/ticket/26838#comment:8
15:04:29 <irl> i just commented on there
15:04:34 <irl> basically two ux things
15:04:43 <irl> i'd like to do #29560 too
15:04:53 <irl> but i could theoretically do all this and deploy it in an evening
15:05:31 <gaba> you would deploy https://research-staging.torproject.org/ as the new research portal
15:05:35 <irl> yeah
15:05:49 <irl> to do that i go to the directory and type "make production"
15:05:52 <irl> and that is it
15:06:18 <gaba> ja, ok
15:06:21 <karsten> should we track it on our roadmap anyway?
15:06:26 <gaba> karsten: waht do you think?
15:06:29 <gaba> yes, we can add it
15:06:46 <gaba> it is only one day of work I think it is just fine to launch it
15:06:51 <irl> i think we should add it, and i should see how much progress i can make on it today and just deploy it if it's good enough
15:07:08 <irl> antonela already signed off on the ux being good enough if i do those two things
15:07:13 <gaba> then  mike could be updating the research ideas with the stuff for the performance grant from mozilla
15:07:14 <karsten> I haven't looked yet, but I trust irl to do the right thing there.
15:07:27 <gaba> sounds good
15:07:39 <irl> gaba: yeah, i really want to make this live so we can start using it more
15:07:59 * gaba is adding it to the roadmap if you all are ok with it.
15:08:06 <irl> (:
15:08:06 <karsten> sounds good to me.
15:08:45 <karsten> next topic?
15:08:48 <irl> ok
15:08:51 <gaba> ok
15:08:54 <karsten> - sbws bandwidth files update (karsten)
15:09:12 <karsten> I just finished a collector patch that downloads bandwidth files.
15:09:28 <karsten> it requires a patch release of metrics-lib, because we made a mistake there.
15:09:37 <karsten> and it requires review and a collector release.
15:09:54 <irl> is review monday going to be too late?
15:09:57 <karsten> I wonder if I should set it to needs_review and then let it run on my own server for a week.
15:10:04 <karsten> I think it's just fine.
15:10:09 <irl> ok cool
15:10:24 <karsten> I think deploying tomorrow morning and then leaving for a week would be too much of an adventure. ;)
15:10:35 <irl> aha
15:11:17 <karsten> okay.
15:11:29 <karsten> this was mostly an update.
15:11:36 <karsten> next update is:
15:11:37 <karsten> - OnionPerf updates (karsten)
15:12:03 <karsten> we discussed a few improvements there.
15:12:08 <karsten> we = acute and I
15:12:40 <karsten> including better error codes in .tpf and .json output and partial completion timestamps.
15:12:55 <acute> #26673 is now on review
15:12:56 <karsten> but all these things use existing logs and just provide better output when processing them.
15:13:43 <karsten> well, I assume that we can use existing logs. pending confirmation.
15:13:52 <acute> yes, we can
15:13:58 <karsten> okay, cool!
15:14:17 <acute> I suggest we try to do this once #26673 gets merged
15:14:50 <karsten> or maybe we can also include #29787, so that we don't do it twice?
15:15:12 <acute> then we'll have an understanding of what may go wrongt once the new errors codes are supported
15:15:47 <karsten> regarding timing, I won't be able to do any reprocessing next week.
15:15:53 <karsten> I could do it the week after.
15:16:26 <acute> if you're ok with this, I can work on the code to do it
15:16:32 <karsten> yes, sure!
15:16:37 <acute> perfect!
15:16:47 <acute> (I may already have a script in progress...)
15:16:51 <karsten> heh
15:17:11 <karsten> oh, another thing regarding OnionPerf:
15:17:24 <karsten> we should talk about some operational/deployment-related things in 1.5 weeks from now:
15:17:49 <karsten> new tor versions, available disk space, different weights for 50k/1m/5m downloads.
15:18:56 <karsten> and finally, I'll look into making new graphs as soon as logs are reprocessed.
15:19:02 <irl> i was thinking we might upgrade op-ab to the latest onionperf codebase and tor version next week assuming no more urgent tasks come up
15:19:10 <irl> one thing i did notice, we don't have the torperf data in the graphs anymore
15:19:26 <irl> the graphs only start with onionperf instances
15:19:31 <irl> did we do this deliberately?
15:19:36 <acute> sounds good, with regards to different weights for 50k/1m/5m downloads: I'd like to better understand how the the *tpf files produced by OP are used
15:20:04 <karsten> irl: next week for the op-ab upgrade sounds good to me!
15:20:11 <irl> well, also depends on acute's availability
15:20:33 <acute> yep, anytime after Wed
15:20:36 <irl> ok
15:20:50 <karsten> irl: not sure what you mean wrt. removing torperf data.
15:21:11 <irl> the graphs on tor metrics no longer show data 2007-2017
15:21:18 <karsten> ah!
15:21:20 <irl> they only show data from op- instances
15:21:25 <karsten> no, that's a bug then.
15:21:39 <karsten> do you want to create a ticket?
15:21:43 <irl> ok, will do
15:21:50 <karsten> great, thanks!
15:22:02 <karsten> (I might not be able to fix that today/tomorrow.)
15:22:14 <irl> that's ok
15:22:18 <irl> as long as we're tracking it
15:22:26 <karsten> acute: not sure what you mean wrt. usage of .tpf files and weights.
15:22:31 <karsten> irl: ok.
15:22:54 <karsten> acute: my idea was that as soon as we have partial completion timestamps we can make fewer 50k requests and more 1m and 5m requests.
15:23:07 <karsten> because they will contain the same information that is also available in 50k requests.
15:23:08 <acute> ok, so do we need to change things further down the pipeline to make use of the new partial completion times?
15:23:19 <karsten> yes, certainly!
15:23:32 <karsten> ah, maybe we can wait a bit then.
15:23:41 <irl> we should keep https://metrics.torproject.org/reproducible-metrics.html#performance updated
15:23:44 <karsten> like, make new timestamps available, wait a few weeks, then change weights.
15:23:49 <irl> i don't think it knows about v3 onions yet
15:23:50 <acute> so do we want to wait until we implement this
15:24:02 <acute> yes, exactly!
15:24:28 <karsten> acute: we should implement the new timestamps soon, but wait a bit before changing weights.
15:24:40 <karsten> and we should announce the new timestamps as soon as they're available.
15:24:51 <karsten> ugh, no v3 onions in that doc?
15:24:52 <acute> sounds like a good transition plan to me!
15:25:08 <irl> we just say "public" vs "onion" but not v2/3
15:26:02 <irl> acute: did you write text for this already in the project report or did we drop it?
15:26:15 <acute> I remember karsten making changes to support v3 address matching
15:26:26 <irl> it was an appendix in the project report to the funder
15:26:39 <acute> so it's just that this document was not updated
15:26:44 <karsten> acute: exactly.
15:26:45 <acute> yes I did write text for this
15:26:56 <irl> ok so maybe we have a patch for the text already
15:27:44 <karsten> https://gitweb.torproject.org/metrics-web.git/commit/?id=7af8f108136b5bb37a0e88f794ce7154c4660ba9
15:27:49 <karsten> that was the commit.
15:28:25 <karsten> do you want to create a ticket and prepare a patch?
15:28:36 <acute> ok, I'll do a metrics-web patch
15:28:42 <karsten> awesome!
15:29:08 <acute> cool!
15:29:09 <karsten> anything else on the OP topic?
15:29:42 <acute> yes
15:29:55 <acute> wrt #26597
15:30:28 <acute> not sure where to go with this ticket
15:31:21 <karsten> so, regarding the last comment there, did you find that tor uses relays with the Guard flag as first hop?
15:31:33 <acute> ticket assumes that we were not using guards as first hops when building circuits through OP - but we found that we are
15:31:40 <karsten> which wouldn't be exactly the same as using a small set of guards as first hops for all circuits.
15:32:18 <karsten> does this make sense?
15:32:34 <acute> yes, it does use (diverse) relays with the guard flag set - I wrote some code to check this for #29373
15:33:22 <karsten> so, the difference might be that OP is picking a new relay with Guard flag as guard for a new, pre-built circuit.
15:33:43 <karsten> whereas vanilla tor uses a relay it has used before as guard which makes it unnecessary to first connect to it.
15:34:06 <karsten> just a quick thought. I haven't looked at #29373 yet.
15:34:19 <karsten> I'll cc myself on #26597 to see what to do with it.
15:34:30 <karsten> oh, I'm already cc'ed.
15:35:08 <acute> this makes sense, I think that is exactly what happens
15:35:37 <karsten> okay, feel free to comment on the ticket. I'll take a closer look later today or tomorrow morning.
15:35:45 <acute> thank you
15:35:55 <karsten> sure!
15:35:59 <karsten> anything else on OP?
15:36:19 <acute> not from me
15:37:43 <karsten> okay.
15:37:50 <karsten> - only reminder: May 16th 14UTC to talk about funding and nlnet (gaba)
15:37:55 <irl> added to my diary
15:38:00 <gaba> ok
15:38:05 <karsten> that will be a busy day!
15:38:14 <irl> i'm not sure about the direction this is going
15:38:15 <karsten> likewise, noted.
15:38:16 <irl> i've lost track
15:38:25 <irl> i guess i will find out on the call
15:38:31 <gaba> yes :)
15:38:41 <karsten> by the way,
15:38:46 <karsten> tordnsel broke over the weekend.
15:38:54 <irl> but then it fixed itself
15:38:57 <irl> so it's fine (:
15:39:02 <karsten> the active scans were broken for several days.
15:39:13 <karsten> a reboot fixed the issue.
15:39:48 <karsten> apparently nothing else broke due to this.
15:39:49 <gaba> good that it fixed itself but I guess you tracked the problem, right?
15:39:58 <karsten> we have no idea why it broke.
15:40:01 <karsten> or how to find out.
15:40:07 <karsten> it's a black box.
15:40:55 <karsten> anyway.
15:41:00 <karsten> next topic?
15:41:29 <gaba> is there anything in the roadmap that needs to be moved around/updated/deleted that was not the stuff already discussed
15:42:31 <irl> not for me
15:42:36 <karsten> loading...
15:43:13 <karsten> hmm, the roadmap doesn't like me.
15:43:15 <gaba> there is ab ug...
15:43:16 <gaba> bug
15:43:21 <gaba> click on 'all the boards'
15:43:23 <acute> #29374 can be moved to done if it's on review
15:43:24 <gaba> at the top
15:43:27 <gaba> and then in the metrics one
15:43:34 <karsten> oh. works now.
15:43:46 <acute> this did not work for me
15:44:31 <karsten> can you reload and try again?
15:44:33 <gaba> we are starting to text the nextcloud from riseup and we may migrate to that in the future
15:45:31 <acute> sadly it loads then presents a gray screen
15:45:37 <karsten> :(
15:45:58 <acute> but I think that's the only update :)
15:46:01 <karsten> so, my two in-progress cards are still in progress, too.
15:46:44 <gaba> ok
15:47:37 <karsten> that means the roadmap is fine for now?
15:47:48 <gaba> yes
15:47:51 <gaba> seems fine
15:47:59 <gaba> i checked all the other tickets
15:48:15 <gaba> one more thing is to start thinking about sessions for the meeting in july
15:48:41 <gaba> but we can have that specific conversation during the monthly meeting
15:48:52 <karsten> okay.
15:49:12 <irl> i'm still not clear on who we even invited
15:49:36 <gaba> I can looked it up and we can review that list in the monthly meeting if you want
15:50:32 <irl> the expenses deadline is the 11th, so that would be too late to make any changes
15:50:37 <irl> i guess it is already too late though
15:50:48 <irl> but to work out what sessions we want we first need to know who we have
15:51:01 <karsten> does the wiki have a list of people who will be there?
15:51:29 <gaba> i think we still did not organized that part of the wiki
15:51:32 <irl> only a list of people that added themselves
15:51:43 <karsten> ok.
15:52:20 <karsten> should we move this discussion to email?
15:52:27 <irl> that sounds good
15:52:31 <gaba> sounds good
15:52:34 <karsten> great!
15:52:48 <karsten> out of agenda items!
15:53:07 <karsten> no meeting next week, I guess? (unless you want to meet without me, of course.)
15:53:32 <irl> i think no meeting
15:53:38 <gaba> no meeting is fine :)
15:53:48 <gaba> we will have double the week after :P
15:54:05 <acute> hehe
15:54:16 <karsten> heh
15:54:24 <karsten> alright. thanks, everyone!
15:54:33 <karsten> have a good day/evening. bye!
15:54:38 <irl> bye!
15:54:38 <acute> bye!
15:54:45 <gaba> bye!
15:54:45 <karsten> #endmeeting