16:00:15 <onyinyang[mds]> #startmeeting tor anti-censorship meeting
16:00:15 <MeetBot> Meeting started Thu Nov 12 16:00:15 2026 UTC.  The chair is onyinyang[mds]. Information about MeetBot at https://wiki.debian.org/MeetBot.
16:00:15 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:00:15 <onyinyang[mds]> hello everyone!
16:00:15 <onyinyang[mds]> here is our meeting pad: https://pad.riseup.net/p/r.9574e996bb9c0266213d38b91b56c469
16:00:15 <onyinyang[mds]> please let me know if you need a link to have write access the pad.
16:00:18 <cohosh> hi
16:00:26 <meskio[mds]> hello
16:02:31 <onyinyang[mds]> The discussion topic in the pad is from last week, Is there anything new to add there or should I remove it?
16:03:12 <cohosh> i can give a short update if there's nothing else to talk about
16:03:28 <cohosh> i made an issue for it: https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/work_items/40583
16:03:39 <meskio[mds]> I'll be happy to hear if you found something
16:03:55 <onyinyang[mds]> It doesn't seem like there's anything else :)
16:04:03 <cohosh> i think the overloading polls are coming from RU
16:04:33 <cohosh> when an overload happens, polls from every region increase, because if there are no proxies clients everywhere need to try again
16:05:23 <cohosh> but the overwhelming majority of polls are from RU, whereas outside of the spikes, polls from RU are pretty comparable to IR and US
16:06:06 <cohosh> what i can't tell is if this spike is due to an enumeration attack, a DoS attack, or censorship
16:07:16 <meskio[mds]> I just checked and there is not reports from our vantage point in russia for a month, so I can't correlate it
16:07:17 <cohosh> i'm thinking the best way to handle this is some sort of client poll rate limit, but it's difficult to tell what that should look like
16:07:53 <cohosh> do you know why this issue was closed? https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/work_items/40466
16:08:44 <meskio[mds]> I don't recall all the details, I think I was convinced by you or shell that it was not a great idea
16:09:04 <cohosh> we have two potential situations: the overload is caused by honest clients using our implementation, or the overload is a custom client that is either too aggressive or malicious
16:09:44 <meskio[mds]> but can be reopened if you do think it makes sense
16:10:10 <cohosh> ok i might open a more generic issue and leave this closed and reopen if we decide it's a good idea
16:10:27 <meskio[mds]> sounds good
16:10:56 <meskio[mds]> yes, be able to differenciate malicious clients from well behaving ones will be nice
16:11:28 <cohosh> i wouldn't want to implement or deploy something without some lightweight simulations like the ones in https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/work_items/25598#note_3392744
16:13:18 <meskio[mds]> makes sense
16:13:49 <meskio[mds]> maybe something to include in a funded project, but I guess will need some more material to produce a full snowflake project
16:14:00 <cohosh> i might update our research page :)
16:14:17 <cohosh> i've geen getting a lot of emails asking about snowflake research projects lately
16:14:27 <onyinyang[mds]> nice :)
16:14:52 <meskio[mds]> ohh, nice
16:16:24 <cohosh> that's it from me on this topic, thanks for the discussion
16:17:20 <onyinyang[mds]> thanks for the update @irc_cohosh:matrix.debian.social
16:17:30 <onyinyang[mds]> I guess there's nothing else today
16:17:31 <meskio[mds]> thanks for investigating it
16:18:06 <onyinyang[mds]> so everyone can have the rest of the time back and we can reconvene next week :)
16:18:21 <onyinyang[mds]> Same UTC time, same channel hehe
16:18:27 <meskio[mds]> :)
16:18:47 <cohosh> :)
16:18:48 <onyinyang[mds]> #endmeeting