IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#i2p-dev
/2026/05/31
&zzz
+FreefallHeavens
+R4SAS
+RN
+RN_
+StormyCloud
+cims
+dr|z3d
+eche|off
+fa
+mareki2p
+nilbog
+orignal
+postman
+qend-irc2p
+red
+snex
+wodencafe
Arch
Danny
Holmes
Irc2PGuest28384
Irc2PGuest28710
Irc2PGuest3231
Irc2PGuest44044
Irc2PGuest59582
Irc2PGuest71568
Irc2PGuest91241
Onn4l7h
Onn4|7h
Over
SilentWave
Sisyphus
Sleepy
T3s|4_
Zapek
aargh4
acetone_
ahiru
anontor2
calamares
dr4wd3
duanin2
justaperson
leopold
mahlay
n2_
nnm--
not_bob_afk
pinotto
poriori
profetikla
rapidash
rotten_egg
test7363673
uop23ip
urist_
user11
w8rabbit
x74a6
zelgomer
orignal my question is what to put there by default depending on profile
zzz (good morning, on 2nd cup of coffee only, don't trust me)
orignal or specofy param explicitly
orignal 30 Kbs is zero
orignal maybe 30Kbs for 2 and 256Kbs for 1 by default
zzz maybe. I just checked 2 routers with about 900 tunnels each. only one tunnel on each is > 30 KBps
zzz remember this is per-tunnel not total for the pool
orignal yes, I understand
zzz all just guesses right now.
orignal do you have an explicit param name?
zzz it's all just for the first 5 minutes until you have a real bandwidth measurement
zzz no I don't have any config yet, or how it would work. The proposal doesn't discuss it either
onon_ Why do you think circuit switching has given way to packet switching in the Internet?
onon_ You are now trying to do circuit switching in i2p.
onon_ This is wrong. You don't take into account traffic patterns on the network.
zzz you mistake me for somebody else. I'm just finishing implemention of something we agreed to a year and a half ago. Nothing more.
onon_ Opening an HTML page with images creates a short-term burst of traffic, rather than a constant flow at a low speed.
onon_ You are killing the i2p network. I was against this proposal when discussing the proposal. And I'm against it now.
zzz that's true, and the http client-side tunnels are the trickiest to deal with
zzz that seems overly dramatic
orignal well, my most concern is telegram
orignal it produces a lot of constant traffic
zzz constant traffic is easy
orignal so, I just need a param
zzz only one? min? max? average? but I'm worried people will screw it up
orignal like 1Mbs
orignal per tunnel
zzz as m or r or l?
zzz you can do that, but that doesn't give you much DDoS protection or traffic pattern protection though
onon_ It would be better if you do something really necessary and useful.
orignal all I need is less stuck tunnels
onon_ Not this nonsense.
zzz better bandwidth management will lead to less stuck tunnels
onon_ It’s better to fix the disconnects if you have nothing else to do.
zzz we gave you voice here on orignal's endorsement, please reduce the snark
orignal please tell me, if I change nothing how would you handle hiigh-loaded(telegram) tunnels?
zzz they will get a default limit which is currently very high, but less than the total limit for the whole router
orignal what exactly is "very hight"?
zzz this simply prevents one tunnel from using _all_ the bandwidth. that's all. it's not a big change
zzz very preliminary, currently 1/12 of total bandwidth. TBD.
orignal not much
zzz I might change it to a % of available bandwidth instead
zzz maybe. just a placeholder for now
orignal please remind me where do you send this param to make sure I handle that block properly
zzz in the TBM properties, see prop. 158
zzz I've been sending m and r for a couple releases now. I just added l
orignal then it's fine
zzz so you should see a lot of m and r
orignal I think I just ignore that block
zzz sure, if you had a problem you'd already know it
zzz just checked, I've been sending m and r since API 0.9.66 2025-06-02 so exactly a year
orignal great
zzz I've also been replying with 'b' since then, so you've known how much bandwidth you "have" for a year also
orignal still can't find this field
zzz one sec
orignal it doesn't seem in the specs
orignal 56-xtunnel build options (Mapping)
orignal this one?
zzz correct
orignal thanks
orignal will check
zzz I called it "properties" but thats the java class. Our specs call it a "mapping"
orignal just ignore that field for now
orignal I have common code for mapping now
zzz ofc
zzz I think 'l' could be really helpful for mitigating some attacks
zzz m and r and bandwidth management are more complicated and will take some time to figure out
orignal maybe
zzz what do you think, what % of a router's bandwidth should one tunnel be allowed to use?
orignal I would say like 80%
zzz I'm thinking it would be pretty high for low-bw routers, but lower for high-bw routers that may have multiple fast tunnels
zzz also depends if it's a % of total bw or available bw
zzz the problem is if you have one fast tunnel using 80%, and then it gets rebuilt thru the same peer. How do you know whether to accept it or not
zzz not easy to figure out
orignal I would accept
zzz this is not about making all tunnels slow. It's about making fast tunnels fast. It will really help telegram tunnels if we do it right
StormyCloud zzz dev fleet updated to -14. Only warning messages I see are 'WARN [le Jetty-127] rg.eclipse.jetty.server.Server: /jsonrpc/: java.lang.NullPointerException: Cannot read the array length because "addr" is null' but its not on every host so could be isolated issue.
zzz you're using i2pcontrol?
zzz ok. if you want you can give it to the new guy for his first real production bug
zzz if there's no stack trace in the logs, you or he will need to set /configlogging org.eclipse.jetty.server.Server=WARN and wait for it to happen again to get the trace
zzz or, if you would like me to chase it, I'll need the stack trace from you first
zzz may not necessarily be a i2pcontrol bug, maybe something elsewhere, esp. if it's only on the dev builds
zzz good stuff, let me know how you want to play it
StormyCloud no stack trace, I added the override. Ill monitor to see if it produces anything
zzz super, thanks
orignal StormyCloud what's your stat about telegram?
orignal zzz, if I don't request bandwicth do you reply with b?
StormyCloud orignal looks like 2000 active connections if prometheus scrapper is correct.
orignal can you see streams on the router?
StormyCloud I have no idea how to see that.
zzz orignal, no, spec says not required
orignal would be nice to do it
orignal I create a tunnel
orignal than take min from all hops
zzz orignal, if you include any of m/r/l, you will get back a b, then you can calculate minimum b
orignal I know, just curious
zzz StormyCloud, java console does not provide any visibility on streams, except for i2psnark
orignal how can you find out number of streams per dest?
zzz you can't unfortunately
zzz you can get an estimate of recently-connected peers by counting leasesets for the dest on the leasesets(client) tab or counting tagsets on the debug session key manager tab
orignal then nevermind