15:59:27 <GeKo> #startmeeting network-health 05/30/2022
15:59:28 <MeetBot> Meeting started Mon May 30 15:59:27 2022 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:28 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:59:40 <GeKo> okay, let's do some network-health/metrics sync
15:59:49 <GeKo> let's see who is around today
15:59:53 <GeKo> i know ggus is out
15:59:55 <juga> o/
16:00:00 <hiro> o/
16:00:07 <GeKo> aha, hi!
16:00:15 <GeKo> http://kfahv6wfkbezjyg4r6mlhpmieydbebr5vkok5r34ya464gqz6c44bnyd.onion/p/tor-nethealthteam-2021.1-keep
16:00:18 <GeKo> is our pad
16:03:57 <GeKo> okay. nothing is marked as bold
16:04:16 <GeKo> it seems there is nothing you need any help with, juga and hiro?
16:04:34 <juga> GeKo: not at the moment
16:04:34 <hiro> uhm I am not sure you know
16:05:01 <hiro> but I was looking at the spikes I have in https://metrics.hiro.network/userstats-bridge-transport.html
16:05:59 <hiro> and I was looking at the ratio of flakeys vs metrics.tpo
16:06:16 <hiro> and this is approx 2
16:06:23 <hiro> although we have more than 4 flakeys
16:06:48 <hiro> so I was wondering if we approx know how many users does each flakey serve approx
16:07:08 <hiro> but I guess that's part of the problem we are trying to solve
16:07:42 <hiro> this is regarding https://gitlab.torproject.org/tpo/network-health/metrics/website/-/issues/40047
16:07:49 <GeKo> sounds a bit like https://gitlab.torproject.org/tpo/network-health/metrics/analysis/-/issues/40012 could be related?
16:07:54 <GeKo> yeah
16:08:12 <hiro> yeah I was documenting the code we are using in the wiki
16:08:30 <hiro> because of the logic leaves in the DB and some in the code
16:08:37 <hiro> s/leaves/lives
16:08:43 <GeKo> good idea
16:08:52 <hiro> ok I will ask in the ticket maybe
16:09:02 <GeKo> i am worried about the behavior with the spike you found when testing your code
16:09:08 <GeKo> sounds good, too
16:09:13 <hiro> yes although I didn't change any of the logic
16:09:22 <GeKo> which is why i marked the patch as needs'revision
16:09:24 <GeKo> https://gitlab.torproject.org/tpo/network-health/metrics/website/-/merge_requests/20#note_2808465
16:09:34 <hiro> just added the nickname to the set of columns which identify a node
16:09:36 <GeKo> do you plan to dig deeper into that hole?
16:10:36 <hiro> uhm I think I will certainly try to document all we have in the code
16:11:02 <hiro> because we need to have the same logic when we migrate to a DB based pipeline
16:11:12 <GeKo> right
16:11:32 <hiro> anyways
16:12:29 <GeKo> okay, we'll see whether we find something useful for #40047 in that process i guess
16:12:32 <GeKo> thanks
16:13:27 <GeKo> alright
16:13:50 <GeKo> i think the overloaded guards someone pasted on the pad are not problematic right now
16:14:05 <GeKo> however, when looking at the graphs this morning i saw a jump in exit node overload
16:14:27 <GeKo> so i guess congestion control is (finally) kicking in with the tor update in tor browser
16:14:37 <GeKo> i'll keep a closer eye on that
16:15:06 <GeKo> otherwise nothing to report here from my side (apart from the things i mentioned on the pad)
16:15:32 <GeKo> i guess that means we can close the meeting and get back to do something even more useful :)
16:15:42 <GeKo> thanks everyone and have a nice week!
16:15:44 <GeKo> #endmeeting