16:01:09 <GeKo> #startmeeting network-health 09/27/2021
16:01:09 <MeetBot> Meeting started Mon Sep 27 16:01:09 2021 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:09 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:01:12 <GeKo> okay
16:01:15 <GeKo> let's start
16:01:21 <GeKo> http://kfahv6wfkbezjyg4r6mlhpmieydbebr5vkok5r34ya464gqz6c44bnyd.onion/p/tor-nethealthteam-2021.1-keep
16:01:27 <gaba> o/
16:01:44 <GeKo> is the pad, please add your items in case you have not done so
16:01:58 <GeKo> and mark those you want to talk about in bold
16:02:27 <hiro> o/
16:03:08 <GeKo> ggus: ping for you in case you are around
16:06:10 <GeKo> okay
16:06:14 <GeKo> let's start
16:06:43 <GeKo> i have one discussion item
16:07:00 <hiro> ok
16:07:03 <GeKo> and if ggus still shows up i can chat a b it about the eol work i did so far
16:07:15 <GeKo> hiro: i guess we should amend the support article?
16:07:26 <GeKo> and maybe write a follow-up mail to tor-relays@?
16:07:46 <GeKo> given that operators started to show up on irc and wondered why the metrics port did not show them anything
16:07:46 <ggus> o/
16:07:51 <GeKo> hihi
16:07:55 <hiro> yes I did a fix in the code too
16:08:02 <ggus> i will need to leave in 20 minutes.
16:08:08 <hiro> it was a null comparison that was probably causing the issue some people mentioned
16:08:23 <GeKo> ?
16:08:26 <hiro> I was waiting to see if that happened again to make a new release
16:08:50 <GeKo> are we talking about the same thing? :)
16:09:14 <hiro> nope, but I meant that could go in the follow up
16:09:19 <hiro> :)
16:09:29 <GeKo> hah, okay :)
16:09:34 <GeKo> yeah, sounds good
16:09:44 <GeKo> would you mind picking all of that up?
16:10:02 <GeKo> that is support article update and tor-relays@ follow-up?
16:10:59 <hiro> yes
16:11:08 <hiro> sounds good I'll add that to my tasks
16:11:48 <GeKo> thanks
16:12:05 <GeKo> do we have anything else we should discuss around the overload topic?
16:12:16 <hiro> I don't think so
16:13:47 <GeKo> okay
16:14:13 <GeKo> ggus: so, i cleaned up the eol policy or what could be our start
16:14:15 <GeKo> https://gitlab.torproject.org/tpo/network-health/team/-/wikis/Relay-EOL-policy
16:14:31 <GeKo> and i started to create a spreadsheet with eol relays that have contact info
16:14:45 <GeKo> in the first round that gave me ca. 390 relays
16:15:08 <GeKo> but that operated on yesterday's data i had in my bad-relays database
16:15:22 <GeKo> i am not too happy with that
16:15:42 <GeKo> so i'll do something differently tomorrow just fetching onionoo data
16:15:49 <GeKo> and generate the csv file out of that
16:16:04 <GeKo> ggus: should we keep that on in nextcloud?
16:16:11 <ggus> yes, gimme access
16:16:25 <GeKo> i do, once i have the one i am happy with
16:16:26 <ggus> GeKo: can you list all the tor relay versions that are EOL for me?
16:16:36 <ggus> in a ticket
16:16:37 <GeKo> should be tomorrow
16:16:40 <ggus> ack
16:17:13 <GeKo> yeah, i can add that to network-health#56
16:17:18 <GeKo> err
16:17:23 <GeKo> network-health/team#56
16:17:24 <ggus> i want to use metabase+onionoo to fetch that list
16:18:04 <GeKo> you mean you want to create that spreadsheet yourself?
16:18:11 <GeKo> or what is the idea?
16:18:58 <ggus> no no, i just want play around with metabase+onionoo to see if can be useful next time
16:19:05 <GeKo> aha
16:19:17 <hiro> I can also give you a grafana dashboard
16:19:18 <GeKo> well, i'll use onionoo, so that's that
16:19:43 <hiro> with a interactive table that you can use to search and filter
16:19:43 <GeKo> but i'll add to the ticket what i am checking this time
16:20:29 <GeKo> ggus: am wondering about a general template for the mails we send
16:20:35 <GeKo> i guess we should create one?
16:21:25 <ggus> GeKo: yes, we can recycle your previous EOL email
16:21:37 <GeKo> aha, right, good call
16:21:41 <GeKo> i'll go over that
16:21:55 <GeKo> and then add what i still like to the wiki page
16:22:22 <ggus> +1 wfm
16:22:27 <ggus> for EOL bridges, i would like to add something about enabling OBFS4.
16:22:44 <GeKo> well
16:22:59 <GeKo> the problem with bridges right now is that we don't have the contact info
16:23:23 <ggus> yeah, i was thinking to open a ticket and restart the discussion about bridges and contact-info
16:23:26 <GeKo> i guess we could put that ticket onto our roadmap if we think that's useful
16:23:47 <hiro> what was the rationale behind removing the contact-info?
16:23:49 <GeKo> just use https://gitlab.torproject.org/tpo/network-health/metrics/collector/-/issues/20983?
16:24:13 <ggus> GeKo: ty!
16:24:22 <hiro> ok thanks GeKo reading that
16:24:57 <GeKo> the ticket and lnks there is all i know
16:25:25 <GeKo> i *think* the reasoning was erring on potential issues for bridge operators
16:25:41 <GeKo> however, to my knowledge those issues were never spelled out
16:26:10 <GeKo> okay, that's all i had on my plate to talk about
16:26:26 <GeKo> do we have anything else for this week?
16:26:57 <ggus> i will submit a new project to Outreachy
16:27:02 <ggus> about gamification
16:27:12 <ggus> i will share the pad with you, GeKo
16:27:26 <GeKo> great, please do
16:27:54 <GeKo> i'd started reading some papers about that more or less recently
16:28:10 <GeKo> as i think this could be a really cool and useful thing to get going
16:28:16 <GeKo> so, exciting :)
16:29:42 <hiro> I just wanted to mention I think I have digged out the issue with collector and will probably write a proposal to modify how we download descriptors from dirauths
16:29:57 <GeKo> aha!
16:30:01 <GeKo> nice
16:30:11 <hiro> yeah spent some time with collector in debug mode during the weekend
16:30:14 <ggus> i need to go. o/
16:30:29 <GeKo> is that related to the bw gaps we see on metrics, too?
16:30:32 <GeKo> ggus: o/
16:30:54 <hiro> some dirauths are just interrupting the download or refusing the connection, so then collector tries another dirauth and that might not have the descriptors it asks... this much I have uderstood.
16:31:10 <hiro> GeKo: I have to see how that graph is built on metrics-web side in the r-server code
16:31:26 <GeKo> okay. one bug at a time :)
16:31:36 <hiro> more likely the r-server reads a csv and that csv has holes in it
16:32:03 <hiro> so I guess metrics web reads the bandwidth data and decides not to publish it for that day... I am not sure because there is a check about missing data
16:33:17 <GeKo> k
16:33:24 <GeKo> we'll see i guess
16:33:42 <GeKo> alright, let's call the meeting then
16:33:49 <hiro> great! talk later
16:33:51 <GeKo> have a nice week everyone o/
16:33:53 <GeKo> #endmeeting