eyedeekay
Nobody push to gitlab for about 5 minutes.
zzz
0) Hi
zzz
hi
zlatinb
hi
orignal
hi
zzz
what's on the agenda for today?
orignal
SSU2 introducers
eyedeekay
hi
orignal
and release results
zzz
ok, 1) is SSU2 introducers
zzz
2) is release results
zzz
I'll add 3) SSU2 development status
zzz
anything else?
zlatinb
yes, startup expl. tunnel build optimizations
zlatinb
just want to check how i2pd does it really
zzz
ok that's 4)
zzz
one sec while I write it all down
zzz
1) SSU2 introducers
orignal
so as I see we bulish the same keys but iHost and iPort
orignal
because we want to take it from RI
zzz
I'm already lost :(
orignal
what fields do we publish in SSU2 address interoducers?
orignal
iTag, iExp, iKey (or iIdent)
orignal
and we don't publish iHost and iPort
zzz
right. itag, iexp, and iIdent or maybe call it iRouterHash for now
zzz
right, no ihost, no iport
zzz
that's the proposal from last week
orignal
yes, that's what I meant
orignal
and we also agreed an RI might have two SSU2
orignal
ipv4 and ipv6
zzz
right
orignal
and we take one from router's caps
orignal
so if we see "4" we must use ipv4 adddress
zzz
I _think_ so, but I haven't really thought that far
orignal
I'm trying to write that code
orignal
basically data structures
orignal
beside it everything is clear
orignal
so should we leave iKey or use another name?
zzz
ok, but my head is full of "phase 1" and about to explode, so it's hard for me to help on phase 2 right now
zzz
re: leave Ikey or not, I don't know
orignal
phase2 is fine
orignal
but I need to finish parser
zzz
for now, I suggest you guess what is the best way to do it
orignal
fine
zzz
just remember that router infos are untrusted input from other users, don't ever die on missing/bad values, bad base64, bad lengths, etc.
zzz
anything else on 1) ?
orignal
no
zzz
2) release results
zzz
about 23% of net updated
orignal
do you see improvements?
zzz
doesn't look like biglybt has updated yet
eyedeekay
Anecdotally the Reddit crowd is pretty happy
zzz
I think I see a little improvement, but of course I've been running with fixes for a month
zzz
I just asked on zzz.i2p for feedback for people that did NOT run with any fixes a month ago
zzz
I've had good luck talking to our gitlab in the last week
zzz
eyedeekay, could you elaborate on the reddit consensus?
orignal
improvement about failed tunnels of course
zlatinb
I see big improvement in muwire
eyedeekay
The reddit users who have contacted me have, in particular, noted improved reachability of popular sites, getting to sites in <=1 refreshes
zzz
I don't have any tunnel fail data, but of course I've been running with the fixes for weeks
eyedeekay
Much more acceptable behavior for browsing eepsites in general
orignal
good to know
zlatinb
the big improvement came suddenly when the old i2pds crashed
orignal
that's what I mean
zzz
my expl. build success chart is up from about 43% to 46%, but a little too soon to really know what the average is
orignal
so most of i2pds are new now
zzz
because there's variation on weekday vs. weekend
orignal
Transit: 599.35 GiB (1800.96 KiB/s)
orignal
I see more transit traffic than before and on all my nodes
zzz
bottom line, all seems good so far
zzz
anything else on 2) ?
zzz
hi RN
orignal
not from me
RN
:)
zzz
3) SSU2 development status
zzz
I guess I'll go first
orignal
fine
orignal
and tell me your idea to publish it as SSU ))
zzz
I've checked in about 5000-6000 lines of code and comments and related changes
zzz
since the release
zzz
mostly untested and not hooked in yet
zzz
the changes broke my test code in a few places and I'm fixing those now
zzz
the bulk of the changes are now in and within a few days I'll be able to start testing on a testnet
zzz
and fixing what's probably dozens of bugs
zzz
I'm still ready to do testing with test code vs. i2pd whenever you're ready
orignal
I'm not ready yet
zzz
hope to have things working in the testnet in maybe a month
orignal
so tell me how do I recognize if SSU address is SSU2?
zzz
so, same as for NTCP: it will have a v=2, a s=.... and (if inbound) a i=...
orignal
got it
zzz
that's the end of my status, orignal how are you doing?
orignal
so, I have implemented parsing of incoming packets
orignal
first I try to decrypt first 8 bytes
orignal
I calculate key using my own intro key for that address
orignal
then I try to find session by connID in the table
zzz
great, good start
orignal
if it's not there I try to find it amoung pending session requests by endpoint where a packet comes from
orignal
if still not found I assume it's new incming session
orignal
that's all for now
zzz
super, thanks for the update
zzz
anything else on 3) ?
orignal
from another side I develop code to support SSU2 address
orignal
in my RI
orignal
so I will start publishing it sson
orignal
as SSU2 of course ))
zzz
ok
orignal
that's all
zzz
4) startup expl. tunnel build optimizations
zlatinb
yes
zlatinb
so I did some analysis and 75% of exploratory tunnels build in less than 500ms
zlatinb
and the longest successful was about 2500ms
zlatinb
right now java sets the timeout to 13 seconds. What is the timeout for tunnel build for i2pd?
orignal
let me check
orignal
const int TUNNEL_CREATION_TIMEOUT = 30; // 30 seconds
orignal
so I delete pending tunnel after 30 sec
zlatinb
how many exploratory tunnels are built in parallel on startup?
orignal
depedning on config
orignal
3 by defualt
orignal
for each side
zlatinb
and how many tunnel build attempts are allowed in parallel?
orignal
all tunnels of exploratory?
zlatinb
exploratory
orignal
as many as I need
orignal
usually 3
orignal
at the beginning
zlatinb
ok thanks
zlatinb
that's all from me, was just curious
zzz
ok sounds like that's it for 4)
zzz
anything else for the meeting?
orignal
I also build cleint tunnels
orignal
but each pool builds tunnels in random intervals
orignal
to avoid spikes
orignal
no
zzz
didn't mean to cut you off, feel free to keep going
zzz
but that's it for me
orignal
if zlatinb has more questions
zzz
thanks everybody
zlatinb
no more questions thanks
orignal
I hope nobody is going to ban Russian IP addresses?>
orignal
since it's new hysteria
eyedeekay
That would probably be self-defeating wouldn't it?
orignal
I saw one idiot who banned Russian IPs on his yddsrasil node
orignal
seriously
eyedeekay
Bummer. Nobody with a brain has a problem with regular Russians.
eyedeekay
This is clearly on Putin.