&zzz
+FreefallHeavens
+R4SAS
+RN_
+Romster
+T3s|4
+acetone
+eche|off
+mareki2p
+nilbog
+nyaa2pguy
+orignal
+postman
+qend-irc2p
+wodencafe
Arch
Daddy_I2P
Danny
Irc2PGuest15434
Irc2PGuest16019
Irc2PGuest20377
Irc2PGuest31938
Irc2PGuest35412
Irc2PGuest48148
Irc2PGuest77921
MatrixBot
NiceBoat
Onn4l7h
Onn4|7h
Over
Sisyphus
Teeed
aargh4
ahiru
ananas
anontor
b3t4f4c3__
cims
dr4wd3_
duanin2
ice_juice
mahlay
makoto
marek
n2
noidea
not_bob_afk2
o3d3
poriori
profetikla
r00tobo
rapidash
rednode
solidx66
stormycloud[m]
sublimia
test7363673
uop23ip
urist_
vivid_reader56
x74a6
zelgomer
zzz
good morning
dr|z3d
for some definition of good. :)
zzz
big improvement in build success overnight, not sure if due to the i2pd release or something else
zzz
nice to wake up to graphs going up instead of down down down
zzz
It may also just be that we popped out of congestion collapse. Hard to say. Both attack groups appear to be still running.
orignal
are you releasing?
orignal
zzz, i2pd was released yeaterday tonight but major changes was done yersterday morning and people often build from trunk
zzz
we should be on schedule
orignal
fair enough
zzz
I see about 3% 0.9.68
nilbog
the debian build unfortunately still has the same error zzz, tried -11-rc (+ the ppa version) and it's the same
orignal
I haven't updated my routers yet
orignal
to 9.9.68 explicitly
zzz
probably not enough for a huge improvement, that's why I think it's probably just getting out of congestion collapse
zzz
locally I'm up from 20% to 75% client build success. expl. from 8% to 18%. net avg expl. now at 28%
zzz
nilbog, the PERCENT error?
nilbog
zzz, yes it's still here in my tests, maybe an issue on my end? i would've guessed that any dependency issue would show when compiling
zzz
don't know. I'll retest here, maybe I fixed it and then broke it
nilbog
ok thanks
zzz
nilbog, reproduced here, investigating, thanks for the report
not_bob
zzz: I'm running a version of i2pd I compiled about 12 hours ago, and the latest i2p+ and things have been somewhat better.
not_bob
Still getting a very large percent of transient tunnels on my flodfills though.
zzz
yup, thanks for the report
zzz
found the problem nilbog, might be a little messy to fix, the resolute/trixie release deb may be delayed
zzz
you should be able use the jammy package instead, or I'll let you know when I have a fix
orignal
I think most of yeatrday change is still on 0.9.67
zzz
nilbog, I've concluded that debian's libjetty12 package is borked and we're not going to be able to use it
zzz
we're going to use the jammy build for everything including resolute/trixie/stable/sid
orignal
zzz I see 8K floodfiils now
zzz
yeah lets see what the botnets are up to
orignal
are they real or not?
orignal
e.g. do they repspond to requests?
zzz
looks like a big new batch of XfG routers, in the US now
orignal
are they good of fake?
orignal
*or
zzz
yeah I'm connected to lots of them
orignal
did you try to lookup?
orignal
I'm wondering if they respond
zzz
havent done any testing of them, no. A lot are so new they're reporting netdb.knownLeaseSets = 0
zzz
back shortly
dr|z3d
floodfills look normal from here.
dr|z3d
I suspect the botnet ones are swallowing lookups.
dr|z3d
so botnet appears large, >10K large.
orignal
10K FFs?
orignal
are they G?
dr|z3d
check
orignal
I'm busy
orignal
10K routers itself is not a problem
dr|z3d
*** chuckles. ***
dr|z3d
no, I mean "check" as in "yes".
orignal
so 10K G routers?
dr|z3d
correctomundo.
orignal
then that's the problem from one hand
dr|z3d
of the X flava.
orignal
while G means it's a tunnel originator
orignal
the fix is easy
orignal
limit number of TBR from G
orignal
guys, can we collect ALL their IPs this way?
dr|z3d
nuke those, things look normal again, ff size wise.
dr|z3d
or just take a napalm torch to XG.
dr|z3d
if I gave you a list of 10K hashes, any good?
orignal
XG might mean someone is shutting down
dr|z3d
no, it doesn't.
dr|z3d
not in this context.
orignal
in general
dr|z3d
what it means is a shit ton of abusive routers on fast connections that are intent on causing havoc.
dr|z3d
let's talk specifics, you can present your post-doctoral thesis later.
dr|z3d
I have a sorted list of close to 12K router hashes here. they're not all shutting down.
dr|z3d
and those have all been seen in the last 90m.
orignal
so some of them are not floodfills?
dr|z3d
correct.
orignal
what I don't understand why tunnels are not being built?
orignal
they are G
orignal
nobody is trying to build tunnels through them
orignal
do you guys throttle TBMs too much?
dr|z3d
we throttle requests, sure.
orignal
how much?
orignal
27 per 100 secs max for entire routers?
dr|z3d
and no one would be trying to build requests through them, because G. they're not allowing it. that's the point.
dr|z3d
it's both per router, with a growth factor to spot bursts.
orignal
you ban other ruters
dr|z3d
I'm also watching for bursts, there are some pretty demanding routers out there that are evidentially abusive.
orignal
tell me why I have low rate now
orignal
do you know if they build 1 hop tunnels only?
dr|z3d
you have a low rate bacause you have an allergy to throttling.
dr|z3d
but that's ok, because you've always told us you don't want to throttle. so carry on :)
orignal
I have good rates before this routers
orignal
if they build 1 hop tunnels they should afffect only me
orignal
do you get it?
orignal
looks like the problem is that you trottle entire numeber of TBMs
orignal
dr|z3d yes or no?
orignal
guys please dp something with it for the release
orignal
right now Java routers cause low rate
zzz
reminder: Prop. 169 review tomorrow #ls2 8 PM UTC. recommend being on the echelon server side to avoid netsplits
zzz
^^ dr|z3d orignal eyedeekay
zzz
orignal, we already have XfG mitigations in for the release. Today is just a new batch of them
orignal
yes
orignal
zzz do you trottle TBMs for entire router?
orignal
in the new release
orignal
meaning that it affects everybody, not only abusers
orignal
12K G routers is not a problem
orignal
low rate cause by your throttle is a problem
orignal
btw, let's collect their destinations
orignal
assuming thier tunnels are 1 hop
zzz
orignal, we have both global and from/to TBM throttling. We did not make any changes to it in this release
orignal
but now I bet you hot global limits
orignal
btw, this new fleet respects congesion caps
orignal
if yu are E they stop attacking you
orignal
zzz when you throttle globaly do you raise E or G flag?
dr|z3d
> to answer your question, orignal, from what I'm seeing here, the majority are now f.
dr|z3d
> my guess is that this part of the attack floods the network with un-responsive floodfills which cause poor lookup rates all over.
orignal
dr|z3d do you respond to lookups or not?
orignal
can you confirem that they are unreposnsible?
orignal
I don't see any LS lookup problems
orignal
the rate I'm tallking about is tunnel build success rate
dr|z3d
dunno if they are unresponsive, they're immediately shit-listed.
dr|z3d
over time I do see profiled floodfills with 100% failure rates, though.
dr|z3d
what's your success avg looking like?
orignal
haven't tried
dr|z3d
tunnel build success...
orignal
if one hope I send TBMR instead TBR to the next router
orignal
but when I try to build my own tunnel you drop mine because your global throttling
orignal
even if you received nothing from me before