IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#saltr
/2024/11/15
~dr|z3d
@RN
@RN_
@StormyCloud
@T3s|4
@T3s|4_
@eyedeekay
@not_bob
@orignal
@postman
@zzz
%Liorar
+FreefallHeavens
+RTP
+Xeha
+bak83
+cancername
+cumlord
+hk
+poriori
+profetikla
+uop23ip
Arch
DeltaOreo
FreeRider
Irc2PGuest28511
Irc2PGuest64530
Irc2PGuest75862
Meow
Nausicaa
Onn4l7h
Onn4|7h
Over1
acetone_
anon4
anu
boonst
mareki2pb
plap
shiver_1
simprelay
solidx66
thetia
u5657
weko_
dr|z3d which, in theory, means they'll mostly try NTCP first and only use SSU2 inbound once you've exhausted your NTCP slots. If I'm wrong, I'm sure zzz will let me know :)
uop23ip ok, thanks. i thought it would be the opposite, ssu2 newer/better/lighter and more used with new routers :)
dr|z3d all about the "cost" of connection to a specific transport.
dr|z3d if you hover over the NTCP2 / SSU2 labels in your netdb section where routerinfos are displayed, it should show you the cost of the transport.
orignal peers prefer NTCP2 because there is an issue with SSU2
orignal i2pd connects through NTCP2 to a new routers
orignal and uses SSU2 only is router has proven itself before
orignal sorry guys I have decided to give up with this issue
RN is SSU2 that unreliable?
orignal RN it can be not that router your are connecting to
orignal you are connecting to router with ident xxx, xxx in RI copies SSU2 address from router yyy
orignal you think you connect to xxx but actually you connect to yyy
RN oh, because of attack attempts?
orignal yes, it's a big hole in the protocol
orignal that's why i2pd connects to xxx through NTCP2 first
RN yeah, I think I recall a bit of discussion about this
orignal and one it see that xxx publishes real IPs can use SSU2
orignal yes, it was a long dicussion
RN does the ntcp2 connection give the ssu2 address?
orignal but since Java huys doens't want to do anything with it for undisclosed reason I gave up
RN I mean, by using ntcp2 can you get the ssu address from the target router?
orignal when you connect through NTCP2 you cannt's connect to anpther router
orignal because identhash is part of SessionRequest
RN right, so you know you have the right router, can you ask it through ntcp2 to give you its ssu2?
orignal when I sucessuffylly connect though NTCP2 I confirm that a router is real
RN instead of what is published?
orignal I don't need to
orignal I see bpth addresses in RI
RN ahhh.
orignal the attack was that a router I was trying to connect to never existed
orignal they just used IP from real router
orignal however if I can connect through NTCP2 it means it's real
orignal that's the answer why more NTCP2 than SSU2
orignal also i2pd doesn't care about cost
RN ok, that makes sense
orignal why Java guys doesn't want to fix this problem it's mistery for me
orignal although it could be fixed many months ago
orignal as zzz suggested to add one more round to MixHash with remote ident hash
orignal and that's all
orignal futhermore NTCP2 has too much noticable overhead now
orignal say my NTCP2 thread consume 2 times more CPU than SSU2
uop23ip mmmm, i got 3k+ transit tunnels, wouldn't i at least get some ssu2 inbounds and not 0?