15:59:16 <GeKo> #startmeeting network-health 06/12/2023
15:59:16 <MeetBot> Meeting started Mon Jun 12 15:59:16 2023 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:16 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:59:22 <GeKo> alright, let's do it
15:59:29 <juga> o/
15:59:37 <GeKo> https://pad.riseup.net/p/tor-nethealthteam-2023-keep is our pad
15:59:56 <GeKo> just as a note: we'll skip the s112 stuff today as we have our bbb sync tomorrow
16:00:00 <GeKo> at 1800 utc
16:00:08 <GeKo> juga: hiro: ggus: ^
16:00:09 <ggus> hi!
16:00:13 <juga> ack
16:00:19 <GeKo> so what else did we have...
16:00:53 <mattrighetti[m]> Hola
16:00:54 <GeKo> one item i had i already covered
16:00:59 <mattrighetti[m]> o/
16:01:02 <hiro> o.
16:01:04 <hiro> o/
16:01:12 <hiro> I need to fill the pad
16:01:21 <hiro> been fighting a little bug on victoriametrics
16:01:22 <GeKo> ggus: hiro: it seems we are on the hook for an all hands session next week on wed
16:01:28 <RishadBaniya[m]> hey there, i hope i got in time
16:01:31 <GeKo> do i see that right?
16:01:39 <juga> RishadBaniya[m]: yup
16:01:39 <GeKo> RishadBaniya[m]: you did, all is good
16:01:43 <mattrighetti[m]> Will update the pad later
16:01:43 <hiro> but most of my week would be to work on the metricsdb-01.tpo deployment
16:01:47 <hiro> GeKo already?
16:01:56 <hiro> I thought we had more time for a show and tell
16:02:06 <GeKo> 2023-06-21: Combating Malicious Relays, presenters Georg, Hiro, Gus
16:02:34 <GeKo> i mean we can talk to pavel if that's unrealistic
16:02:57 <GeKo> but if not we should likely get together some time this week to do some brainstorming maybe?
16:04:03 <hiro> uhm ok I think it is doable but would have preferred more time
16:05:12 <GeKo> ggus: what do you think?
16:06:11 <GeKo> while ggus is pondering we can move on
16:06:27 <ggus> more time would be better
16:06:28 <hiro> the reason is so that I can get the deployment this week without thinking about the presentation
16:06:58 <GeKo> okay
16:07:09 <GeKo> i can mail pavel so we can move things around
16:07:19 <GeKo> 2 weeks later?
16:07:53 <GeKo> which would be july 5
16:08:02 <ggus> wfm
16:08:23 <hiro> ok! sounds good
16:08:43 <GeKo> great
16:08:46 <RishadBaniya[m]> idk if this is the right place to ask the question, but are these constraints enough to consider while creating circuits for partition checking tool https://gitlab.torproject.org/tpo/core/torspec/-/blob/main/path-spec.txt#L303
16:09:58 <GeKo> RishadBaniya[m]: i think the first "-" items are good
16:10:20 <juga> agree
16:10:55 <GeKo> any other questions/topics to talk about?
16:11:04 <GeKo> how is gsoc going in general?
16:11:12 <GeKo> mattrighetti[m]: RishadBaniya[m]: you need help with anything?
16:11:44 <mattrighetti[m]> All good here
16:12:25 <GeKo> great
16:12:37 <mattrighetti[m]> Will have some endpoints ready to test by the end of the week hopefully
16:12:46 <RishadBaniya[m]> well, tons of time spent on peeling abstractions
16:12:52 <RishadBaniya[m]> it's fun tbh
16:13:13 <GeKo> nice
16:13:40 <RishadBaniya[m]> i wanted to know your thoughts on using graph database for storing the output
16:14:06 <juga> RishadBaniya[m]: i think that'd be great, but not sure we'll have time
16:14:26 <juga> on a quick search, i also didn't find a rust implementation that looks stable
16:14:46 <juga> maybe we can start with a DBRM, and see later on
16:14:52 <juga> *relational database
16:15:36 <juga> sounds good?
16:15:52 <RishadBaniya[m]> im going all in this week, as i have complete holidays, should i research more?
16:15:59 <RishadBaniya[m]> on what other options
16:16:02 <RishadBaniya[m]> can i explore further?
16:16:18 <juga> ok, dedicate a few hours maybe :)
16:16:47 <juga> you have any opinions on this GeKo?
16:16:50 <GeKo> nope
16:16:52 <GeKo> :)
16:16:59 <juga> fine
16:17:32 <juga> RishadBaniya[m]: i can add in other later, what i found about graph dbs and rust (not much)
16:17:37 <GeKo> ggus: i have one item for you: to check out the thread on dir-auth@
16:17:38 <juga> to an issue/wiki
16:17:51 <GeKo> i think in particular alex's last  mail is relevant...
16:18:02 <GeKo> juga: sounds good, thanks
16:18:29 <ggus> k, i'm still catching up my emails from last week
16:18:34 <GeKo> ggus: which reminds me that we should probably sync some time this week as a bunch of things have piled up i guess
16:18:41 <ggus> yup!
16:19:15 <GeKo> let's do the 2 person doodle outside this meeting...
16:19:22 <GeKo> anything else for today?
16:19:30 * juga is good
16:19:38 <ggus> all good!
16:19:56 <GeKo> reminder for the folks working on sponsor112: we'll have our monthly bbb sync tomorrow 1800 utc
16:20:09 <juga> thanks for the reminder!
16:20:23 <GeKo> hiro: all good?
16:20:38 <hiro> good
16:21:04 <RishadBaniya[m]> i had one question, i almost forgot to add
16:21:23 <GeKo> RishadBaniya[m]: go ahead
16:21:26 <RishadBaniya[m]> nick had mentioned that calling timely_netdir
16:21:29 <RishadBaniya[m]> and storing the output
16:21:31 <RishadBaniya[m]> for long period of time
16:21:36 <RishadBaniya[m]> leads to excessive memory usage
16:21:51 <juga> yup
16:22:01 <RishadBaniya[m]> and this partition checking tool
16:22:06 <RishadBaniya[m]> seems to hold this for long period of time
16:22:10 <RishadBaniya[m]> it seems right?
16:22:24 <RishadBaniya[m]> so how does this affect
16:22:26 <juga> as i interpreted, i think it should be call every hour
16:22:28 <RishadBaniya[m]> the tool that we're building
16:23:00 <juga> or have some event listener (if possible) that get notified on netdir updates
16:23:14 <GeKo> that, yes
16:23:36 <GeKo> or we can just have it run every X hours to begin with
16:23:44 <juga> but i don't know much about arti/rust, so you can try one of those approaches, and we'll see how it performs
16:24:16 <juga> we can show arti people some concrete piece of code too, and ask them what they think
16:24:36 <GeKo> +1
16:24:49 <RishadBaniya[m]> yeah there is a method called events() that returns stream of DirEvent, so as of right now the only way i could find was to get the updated netdir and consider influx of new relays and new descriptors is by listening on that stream and updating our NetDir when we receive some new DirEvent
16:25:06 <mattrighetti[m]> Rishad Baniya: if you need any kind of help with rust you can ping me, I’ll try to help
16:25:32 <RishadBaniya[m]> thanks mattrighetti , sure!
16:25:38 <juga> RishadBaniya[m]: that way looks good to me
16:26:21 <juga> *sounds
16:26:54 <GeKo> okay, folks. i think we can close the meeting for this week. thanks everyone for joining and have a nice week
16:26:58 <GeKo> #endmeeting