IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#i2p-dev
/2024/01/31
tennis2 Did this little icon for i2p called eyeeyepeer, so as not to conflict with the classic itoopie made in inkscape, double-optimised (for some reason one pass doesn't do everything) license GPLv3 licensed in comment
tennis2 Its done in the style of papirus
tennis2 rename file eyeeyepeer_32.svg
tennis2 I will be able to do other sizes (16 18 22 24 28 48 64) if interested
tennis2 did it for the 'snarkdisk' script. Putting in a zenity box asking the user if they want to start i2p/i2psnark after the disk is accessible.
orignal zzz we see somthing strange
orignal a lot of floodfills with old version like 0.9.29
orignal but we agreed long time ago that mininal version for floodfill is 0.9.51
zzz those are Vuze bittorrent clients
zzz our agreement was what floodfills we _use_. I can't reach into old Vuze clients and make them not floodfill :)
orignal yes, we ignore them
orignal what we are investigation why number of floodfills is less thna before
zzz before when?
orignal before the last attack
orignal but why Vuze even turns on floodfill mode?
zzz because it's automatic. we don't want applications to disable floodfill. that's bad
orignal wait. what if I don't want to be a floodfill?
zzz you can disable it manually in the console. but the default is auto
zzz auto == enough bandwidth, etc.
orignal yes, but Vuze is a build-in router
zzz right, then you would have to edit the router.config file
orignal idk how Vuze was developed
orignal so do you see less number of floodfilld now?
orignal I mean elgible floodfills
orignal e.g. >=0.9.51 and reachable by ipv4
orignal we are investiigation what cauuses it
zzz we talked about it before. my ff count is very flat this year. I do see a big drop around the time of our release late Dec.
orignal and we have discovered bunch of old floodfiils
orignal we talked during the attack
orignal and no attack now but number of floodfills is not back
zzz I'm flat between 1100-1400 ffs since Jan. 1 to today (on a non-ff)
zzz want a graph?
orignal we see 700-800
orignal we also drop floodfill if we can't reach thier addresses
orignal looks like you don't filter something that we do
zzz maybe
orignal if we see a router was not reachable we exclude it for 2 hours
orignal if they published IPs
zzz our release rolled out from about 12/18-12/31, and the attack was 12/25-1/24, it's hard to separate out the two things for cause and effect
orignal yes, but we are talking about now
orignal what we check that you don't check
orignal !r->IsFloodfill () || !r->GetProfile ()->IsUnreachable ()
orignal we simply check if floodfill and was not reachable recenttly we exclude it
zzz we do that too, for all routers, not just ff, but only for about 8 minutes. 2 hours is a long time
orignal let me try
zzz or just look and see how many you have banned at once
zzz I currently have 14 routers banned for "unreachable on any transport". don't know how many are ff
orignal Vort is investigating
orignal if you saw a floodfill thaty suddently because U what you do?
zzz I think we would skip it
orignal I mean you consider that U real or as failed peer test?
zzz when we were selecting floodfills for a query or store
orignal so you don't consder U routers as floodfills?
orignal Vort sees that if we add Uf routers to our numbers we receive your numbers
zzz it's complicated
zzz we have known floodfills, unusable floodfills, banned floodfills, and usable floodfills
zzz and in usable we have: good, ok, and bad
zzz based on profiles
zzz actually looks like I was wrong, we don't pay attention to U cap
orignal and what is 1250? all or thme?
orignal but how do you connect to such floodfills? through introducers?
zzz all known in the netdb at the time. everything except banned
zzz same way we connect to any router
orignal but an updated floodfill with U doesn't conting addresses anymore
zzz we don't 'update' RIs like that, changing caps, etc.... it's all done in profiles
zzz RIs are read-only
zzz if we can't connect to it, we mark it in the profile