13:02:46 #startmeeting Network Health 2026-11-02 13:02:46 Meeting started Mon Nov 2 13:02:46 2026 UTC. The chair is hiro. Information about MeetBot at https://wiki.debian.org/MeetBot. 13:02:46 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:02:56 #link https://pad.riseup.net/p/tor-nethealthteam-2026-keep 13:04:01 i guess i can start? 13:04:09 ok GeKo (IRC) it's just me and you I think.. 13:04:22 so, last week was a bit shorter than usual but intense nevertheless 13:04:30 ah, okay, that's fine 13:04:31 Juga mentioned they were solving some paper work issue... 13:04:41 so please go ahead 13:04:59 alright. apart from the bad-relay stuff i worked on two things 13:05:17 the report for p183 is mostly done from my pov 13:05:38 i need to fix small things and fine-tune it a bit but the the meat should be there 13:05:49 nice! thanks! 13:06:00 https://gitlab.torproject.org/tpo/network-health/team/-/work_items/424#note_3482075 13:06:05 is the relevant commit 13:06:32 hiro: you could look over it as i guess we'll talk about it with gabaaa[mds] on wed 13:06:58 if both of you are happy with it i'll just polish it up and we are done on that front 13:07:05 yep yep thanks I am going to do that... 13:07:24 apart from report writing i workeed with netteam folks on the dos and the 0.4.9.15 release 13:07:33 i think the defenses look good so far 13:07:45 there is fallout, though: https://gitlab.torproject.org/tpo/network-health/analysis/-/work_items/116 13:07:53 thanks for that 13:07:59 i suspect that's not due the dos defenses but the TAP rejection 13:08:13 which means 0.4.8.x clients are not allowed in the network anymore 13:08:31 i heard dir-auths are getting hammered now 13:08:42 so, i suspect they will bring this up later today in the sync 13:09:03 but they should be aware of this ticket already (at least a bunch of them) 13:09:23 meaning they try to connect continuously? but is it only dirauth or also fallbacks? 13:09:28 *dirauths 13:09:42 this week, apart from the many meetings, i'll try to investigate this issue and get back to p203 and the data format spec 13:09:57 i need to look at the fallbacks. not sure yet 13:10:23 it could be that the 0.4.8.x ones are so old that essentially only dir-auths are still reachable from the fallback list 13:10:52 it's puzzling that they get counted though 13:11:23 I was looking at: https://grafana2.torproject.org/goto/s728wm?orgId=1 13:11:37 and especially at the tunneled-dl clients count 13:12:06 it resminds me a bit of https://gitlab.torproject.org/tpo/network-health/analysis/-/work_items/101 13:12:23 but the underlying reason for that got fixed 13:13:00 and i was somehow assuming we'd avoid spikes like the ones in #116 with those fixes 13:13:14 and two things struck me. 1. bridge clients also spike and also snowflake, 2. They do not spike for tunneled-dl for bridges but do for relays 13:13:16 but then i've not looked close enough at that ticket yet 13:14:06 that snowflake bw spike probably not relayted 13:14:32 re 1., yeah. i've not looked for 2. but you could add that part to the ticket, if there is anything interesting... 13:14:45 anyway, i think that's all for me for last and this week 13:15:41 yeah I am going to have a look at those thanks 13:17:02 ok I can go, I have pending issues from fixing db data and I am looking back at the notes from the dev meeting and put all the projects ideas into proposals 13:17:19 so that these are ready for next year and also so that the memory is still fresh from what we discussed 13:17:28 will also try to order a bit our priorities on the wiki and so on 13:18:03 I'll cc you all on the proposals created so if you think there is anything to add up you can 13:18:28 sounds good 13:18:40 and that's mainly all for me 13:19:50 something that we can chat about next monday is that maybe once a month we could have a sync by voice 13:20:25 fine with me 13:20:29 but we have to check how to make this work with the various projects syncs... this was an idea from juga so we can chat when they are online 13:21:14 one thing i just realized: metrics-alerts@ is telling is that we start to drop consensuses on collector again 13:21:21 both flavors 13:21:35 did anything change on collector-rs side? 13:21:46 juga: ^ 13:22:46 is there anything I am blogkin you on GeKo (IRC) or anything you need from me? 13:23:32 nope 13:23:34 all good 13:23:57 just the report review so we are set for the wed sync 13:24:50 it does, but I have checked and we are not, unless I am mistaken? 13:25:09 it started with the switch from summer-time to winter-time 13:25:11 or just normal time 13:26:18 I found one missing actually... and this is the first time I do... been checking all week 13:26:22 and it is today at 9am 13:27:13 and another one on the 31st 13:27:23 hrm, okay 13:27:54 so yeah I was mistaken... nothing changed... I only restarted the service 2-3 days ago to update the onionperfs 13:28:34 we miss 2026-11-01-20-00-00-consensus as well it seems 13:29:00 I'll open an issue... 13:29:15 it is weird because we should also have that file locally from moria1 13:29:28 and gabelmoo 13:30:22 and for some notifications it seems the 4h waiting time was not enough 13:30:56 that's for the consensus-ns only messages yesterday at least 13:30:57 ok 13:32:47 yeah it's weird... 13:33:01 anyways I also find it strange that we do not get that from collector-2 13:34:34 It think the 4h waiting is because we got that eventually from collector-2 but not all of them are there... strange that we miss on both services 13:34:44 anywasy I'll open an issue. Thanks 13:35:10 I'll end the meeting then if there isn't anything else 13:35:47 #endmeeting