15:59:01 <GeKo> #startmeeting network-health 10/23/2023
15:59:01 <MeetBot> Meeting started Mon Oct 23 15:59:01 2023 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:01 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:59:04 <mattrighetti[m]> o/
15:59:05 <GeKo> hello!
15:59:06 <juga> \o
15:59:20 <GeKo> let's see what our pad looks like: https://pad.riseup.net/p/tor-nethealthteam-2023-keep
15:59:40 <GeKo> please add your items in case you have not done so yet
15:59:56 <hiro> o/
16:00:19 * juga doing so
16:00:42 <GeKo> hiro: welcome back
16:00:56 <GeKo> i hope you had some time off your computer and could relax :)
16:01:08 <juga> +1 :)
16:02:03 <hiro> thank you
16:03:05 <GeKo> hiro: https://gitlab.torproject.org/tpo/network-health/metrics/descriptorParser/-/issues/52#note_2953681 seems to be a thing juga could need some feedback for
16:03:42 <GeKo> otherwise there is only my afk part marked as bold; we can take that one to the end of this meeting
16:04:46 <juga> yup, we can talk about it later too, hiro
16:05:03 <hiro> I think there have been some issues with the parser
16:05:05 <juga> oh,ic, hiro just replied there
16:05:10 <juga> ic
16:05:24 <hiro> which i hope to fix with the upcoming MR https://gitlab.torproject.org/tpo/network-health/metrics/descriptorParser/-/issues/53
16:05:35 <juga> awesome
16:06:10 <GeKo> so, we could use #52 as a good test case for whether the proposed patch for #53 is actually enough, i guess?
16:06:30 <hiro> possibly
16:06:54 <hiro> the issue is that the parser takes days to complete because I was using the same logic from onionoo to update the family and declared family fields...
16:07:04 <GeKo> or maybe it's something else. okay, sounds promising either way, though
16:07:09 <GeKo> yeah
16:07:12 <hiro> which was go through all the statuses, but this was making the service too slow
16:07:55 <arma2> hello metrics people, i have on my todo list to make a ticket for us to rederive / rediscover how to compute what fraction of tor traffic is onion service traffic. it came up while talking to a researcher about a new paper last week, and it will come up again in the future. i will file this ticket, soon i hope, as the next step.
16:08:16 <GeKo> good!
16:08:26 <hiro> yeah it would be really nice to fix that too
16:08:48 <mattrighetti[m]> hiro: nothing on my part other than those fields that I was talking about previously, I can take care of integrating it in the open ticket
16:08:53 <arma2> especially as i overheard al telling somebody "1%" in the same conversation as i said "3%" and i think the number might actually be "5%" this year :)
16:09:17 <hiro> sure matt that's on my list for the first half of this week
16:09:26 <hiro> thanks!!
16:09:40 <mattrighetti[m]> hiro: thanks!
16:10:02 <GeKo> what are a couple of % among friends anyways ;)
16:10:27 <GeKo> but, yeah, sounds like a good idea
16:10:29 <arma2> geko: that was my reasoning too! but the comms team was not satisfied with it
16:10:36 <GeKo> heh
16:11:24 <GeKo> hiro: we have some pretty deadline things for this week you might want to focus on first
16:11:27 <GeKo> or help with
16:11:36 <GeKo> *pretty important
16:11:45 <hiro> sponsor112? we have also issues on meronense again xD
16:12:01 <GeKo> nothing pretty about those deadadlines
16:12:36 <GeKo> *deadlines
16:12:52 <GeKo> hiro: so for s112 we need the number of tagged relays so far
16:13:08 <GeKo> which is did not get extracted from the db due to permission issues
16:13:25 <hiro> ah ok I can do that
16:13:35 <GeKo> and the info is not on the website available yet
16:13:37 <GeKo> thanks
16:13:50 <GeKo> it would be nice to fix my permission issues, too
16:14:03 <GeKo> i can ping you later for that to get sorted out
16:14:21 <hiro> I think if you log in on the machine as parser you should be able to do it
16:14:32 <GeKo> yeah, no dice, though
16:14:37 <GeKo> i did that :)
16:14:44 <hiro> weird that's what I do
16:14:55 <GeKo> o_O
16:15:50 <GeKo> okay, the other deadline one is we need to assemble again data for s61 indicators
16:15:55 <GeKo> for q3
16:16:04 <GeKo> i am not sure what you did in the past for that
16:16:23 <GeKo> but whatever it was (i think some onionperf related stuff?) we need to do it again :(
16:16:44 <GeKo> it should be the last time we need to do that, though
16:16:48 <GeKo> with one exception
16:17:00 <hiro> uhm
16:17:05 <hiro> ok that sounds good
16:17:13 <GeKo> we need to fix https://gitlab.torproject.org/tpo/network-health/metrics/onionperf/-/issues/40057 and then report data for that in q4
16:17:14 <hiro> except that we have rotated all the onionperf clients
16:17:28 <GeKo> well, just use any data we have for q3
16:17:45 <GeKo> there should be sufficient available
16:17:46 <hiro> ok we might have partials for the tor version 6 series
16:17:49 <hiro> yeah
16:17:54 <hiro> sufficient yeah
16:18:04 <hiro> and should we include conflux results too?
16:18:22 <GeKo> i guess it depends on the indicator
16:18:37 <GeKo> might be smart to ask mikeperry for that as well
16:18:52 <GeKo> as i have not been following that part of s61 closely
16:19:30 <GeKo> yeah, and then there is https://gitlab.torproject.org/tpo/network-health/metrics/website/-/issues/40103 for getting ggus unblocked
16:19:45 <GeKo> which is weird as it's only failing intermittently for months now
16:19:59 <GeKo> and only at the per-country views
16:20:15 <GeKo> while we do have continuous data for the "all" case
16:20:25 <GeKo> (modulo the gaps we know about)
16:20:30 <hiro> yeah
16:20:54 <hiro> I have that on the list... there is nothing in the logs unfortunately
16:21:04 <GeKo> yeah, i saw that
16:21:06 <hiro> I might try to add some indexes to our db
16:21:15 <hiro> maybe that would help speeding things up
16:21:18 <GeKo> and now i am not sure how to debug this further
16:21:21 <GeKo> :(
16:21:37 <GeKo> but the data is not lost, is it?
16:21:40 <hiro> I'll ask anarcat for some help to see if it is the prcess being killed by the kernel again
16:21:52 <GeKo> thanks
16:22:05 <hiro> well I hope we might be able to recover some
16:22:40 <GeKo> okay. we'll see. so, that's all of the pressing things i have for you... welcome back ;)
16:24:33 <GeKo> alright. do we have anything else we need to chat about today?
16:24:40 * juga is good
16:24:55 <mattrighetti[m]> * Matt is good
16:25:08 * hiro is groot
16:25:26 <GeKo> great. so, next monday (in fact the whole week) i'll be afk
16:25:33 <GeKo> what do we wanna do with this sync?
16:25:40 <GeKo> should someone else run it?
16:25:46 <GeKo> should we skip it for that wek?
16:25:49 <GeKo> *week
16:25:59 <juga> hmm
16:26:34 <juga> i can volunteer to run it, though we can also live without it, i guess ;)
16:26:35 <GeKo> i am fine either way :)
16:26:56 <hiro> me too
16:27:05 <hiro> I mean fine if we do it or if we skip
16:27:14 <GeKo> okay
16:27:24 <GeKo> juga: okay, then just run with it next week?
16:27:32 <juga> GeKo: ok
16:27:32 <mattrighetti[m]> We could keep it and move fast if there’s no update
16:27:38 <juga> yup
16:27:40 <GeKo> ggus: you around and have anything to sync about?
16:27:42 <GeKo> +1
16:28:40 <GeKo> does not seem to be the case
16:28:47 <GeKo> thanks everyone then
16:28:51 <GeKo> and have a nice week
16:28:52 <gaba> GeKo: ggus: hiro: at some point we need to meet to discuss https://gitlab.torproject.org/tpo/network-health/metrics/website/-/issues/40103
16:28:52 <GeKo> o/
16:28:59 <gaba> but it could be next meeting
16:29:10 <mattrighetti[m]> Thanks! Ciao o/
16:29:13 <GeKo> gaba: we already did?
16:29:41 <GeKo> see above and on the ticket
16:29:42 <gaba> GeKo: can you update the ticket with any follow up?
16:30:00 <gaba> mmm, it is a memory issue?
16:30:10 <GeKo> likely, we don't know yet
16:30:17 <gaba> anyway, i will comment on the ticket
16:30:25 <GeKo> apart from that i still don't know what ggus actually needs
16:30:42 <GeKo> i flagged that on the ticket a couple of days ago
16:30:52 <GeKo> still waiting on input here
16:30:56 <hiro> if data is needed it could be easier for me to compute it separately by creating the graphs offline
16:31:05 <GeKo> yeah
16:31:20 <GeKo> andway, let's finally close this meeting...
16:31:24 <GeKo> #endmeeting