~dr|z3d
@RN
@RN_
@StormyCloud
@T3s|4_
@eyedeekay
@orignal
@postman
@zzz
%Liorar
+FreefallHeavens
+Xeha
+bak83_
+cumlord
+hk
+poriori
+profetikla
+uop23ip
Arch
DeltaOreo
FreeRider
Irc2PGuest10850
Irc2PGuest19353
Irc2PGuest23854
Irc2PGuest46029
Irc2PGuest48064
Meow
Nausicaa
Onn4l7h
Onn4|7h
Over
acetone_
anon4
anu
boonst
enoxa
glowie
mareki2pb
mittwerk
plap
shiver_
simprelay
solidx66
u5657_1
weko_
orignal
dr|z3d +m wouldn't help if he keeps joining/leaving from hundreds users
orignal
plus flood into PM
orignal
dr|z3d how about simlutenious 3000 connections from own destination each
orignal
I saw like 250K incoming tags
eyedeekay
dr|zed want to take a look at this? git.idk.i2p/i2p-hackers/i2p.i2p/-/merge_requests/218.patch
eyedeekay
Set of tweaks to the netDb page for client Db's, I think it's an improvement
eyedeekay
Not exactly in line with what you proposed
dr|z3d
sure, can take a look, eyedeekay, I've merged some stuff recently, too.
dr|z3d
git.skank.i2p/i2pplus/I2P.Plus/commit/cb5cbd94bc24c2ea4e643076e629b39de348fad0 and git.skank.i2p/i2pplus/I2P.Plus/commit/004d39271cc83db01c8cde6e773a25433739230e (WIP)
RN
yeah orignal, I saw that. the PM's and the nicks each with their own dest. I was wondering if my irc client was going to blow up. LOL
RN
seemed very similar to a floodbot that I saw on an outernet irc not that long ago
dr|z3d
ah, a shitshow then, orignal :)
dr|z3d
those methods mostly work as intende, eyedeekay, except I2PSnark refuses to be detected as local and published for some odd reason. And I'm still trying to figure out how to insert h2s only once for each section for local/clientDb leasesets.
dr|z3d
so my current approach is to have all local leasesets we generate on one page, in sections: server (published), client(unpublished) and requested (non-local services), and leasesets we host when a ff on a separate page only if we're floodfill. I'd also like to get rid of the debug option and assimilate any useful debug info into the main display.
dr|z3d
for non-floodfills, the intention is, once the sections are correctly presented, to have a summary of each section and the leasesets on a display toggle.
eyedeekay
Ok so the floodfill leasesets page, that's easy to hide based on floodfill status, but it might be better to show or hide it base on whether it contains leasesets>0
eyedeekay
Which I will think about
dr|z3d
That's fairly easy, UI-wise.
dr|z3d
All users land on l=3, and if floodfill, a link can be presented at the top of the page (not the nav) to switch to ff context.
eyedeekay
In the case of remote leasesets in subdb's I still don't exactly see the point of separating them from the associated client and it's local leaseset
dr|z3d
that link can indicate the number of ff leasesets.
eyedeekay
Re: debug info, what do you think about an 'also appears in' feature for subdb leasesets, in case they appear in another clientdb or the main db
dr|z3d
If I'm honest, I'm still trying to get my head around clientDb, subDb and the rest of it.
orignal
RN I saw all this ECIES session in the webconsole
dr|z3d
what currently makes sense to me is leasesets I own/generate locally, published and unpublished, and leasesets I've requested based on services I'm accessing.
dr|z3d
tunnel/dest nicknames presented for those where available sheds light on what they're for.
orignal
dr|z3d basically the bot looked like Plaz's bot
orignal
but he swears that it was not him
dr|z3d
can you throttle server joins, orignal?
orignal
yes, we have confirured it
dr|z3d
ok, good. that should help plenty.
orignal
yes
orignal
however it still an issue on i2p side
orignal
many leasets, many sessions many tags
RN
so there's another one besides plaz
orignal
who knows
orignal
I still don't have an explaination why it looked like plaz's bot
dr|z3d
did he publish bot code to github?
orignal
I don't think so
orignal
however he is so mad that he could
dr|z3d
wouldn't surprise me.
eyedeekay
dr|zed the reason that I think client dbs aka subdbs should be associated with certain remote leasesets when we request them is because the important thing about subDb's is that: the application decides when to use a subDb based on the primary hash of the leaseSet of the client used to request the remote leaseSet
orignal
the good thing for this attacke that I have found memory pool that was not cleaned proprely
orignal
fixed
eyedeekay
so every remote leaseSet requested by a client automatically gets bucketed into that client db based on which client requested it
eyedeekay
That way you always know who received a leaseSet so it's not possible for an attacker to send a leaseSet to a client and request it back out via the router and vice-versa
orignal
do you guys have limits for number of ECIES sessions per locat destination?
dr|z3d
yeah, that's backend stuff, the question is how to present it so that the info isn't overwhelming.
orignal
thousands sessions and leasests doesn't sound right
eyedeekay
Before we had a protection against this but it could be gamed by attackers who could wait until they replied with a leaseSet to a client, then send it again to a router with different options, and then request it back out the client, for instance
eyedeekay
there were like, ~4 or 5 attacks like that that subDbs were the solution for
eyedeekay
various patterns of publish-reply-tweak-request
dr|z3d
yup, I'm with you on the reasons behind subdbs, that's fine. it's the UI presentation that needs to be simple for end users, even if that means blurring some of the boundaries, presentation-wise.
dr|z3d
local published, local unpublished, requested remote. if we can tag remote leasesets with the client/nick they're requested from, we've won. we don't need to present them per-client.
eyedeekay
I think the boundaries are pretty important to show in this case, I'm unsure what we're showing without them unless the corresponding client is added to the leaseSet tables when we render and display it
dr|z3d
yup, that.
eyedeekay
Fair enough, that's a valid approach with one advantage, which is that if they're sorted and you have multiple copies in a single db they should appear next to eachother
eyedeekay
so I wouldn't recommend deduplicating
eyedeekay
*multiple copies in multiple dbs
dr|z3d
you can dedupe if you present multiple client nicks/icons in the LS header, perhaps.
dr|z3d
"Requested by: I2PSnark, HTTP Proxy" or somesuch.
dr|z3d
and/or under the leases section, separate per-client leases.
orignal
plaz's bot
dr|z3d
told you, github :)
dr|z3d
he wants bitcoin donations for it as well. the cheek!
uop23ip
Hi orignal, got a problem building i2pd static on arm64. 2 weeks before it worked.Compiler version 12.2.0.:
uop23ip
from /home/i2p/i2pd/i2pd_buildStatic/i2pd/libi2pd/Crypto.cpp:17:
uop23ip
usr/include/boost/asio/awaitable.hpp: In constructor ?boost::asio::awaitable<T, Executor>::awaitable(boost::asio::awaitable<T, Executor>&&)?:
uop23ip
usr/include/boost/asio/awaitable.hpp:68:19: error: ?exchange? is not a member of ?std?; did you mean ?std::__atomic_impl::exchange??
orignal
try to build wjth C++17
orignal
most likely boost issue
orignal
change to c++17 and try
orignal
dr|z3d he is complete idiot then
orignal
uop23ip version of boost please
dr|z3d
*** chuckles ***
dr|z3d
mareki2pb: receiving, over?
uop23ip
Found Boost: /usr/lib/aarch64-linux-gnu/cmake/Boost-1.74.0/BoostConfig.cmake (found version "1.74.0") found components: system filesystem program_options missing components: atomic ,orignal
uop23ip
change to 17 did not work, same error
dr|z3d
tried cleaning your workspace, uop23ip?
dr|z3d
'make clean && make install' or whatever.
uop23ip
always fresh git clone, if you mean that
uop23ip
what about the missing atomic component in boost?
dr|z3d
try: apt install libboost-atomic-dev
Irc2PGuest71319
ouch trying to build a rando C++ program
Irc2PGuest71319
I remember last time I did that. I started drinking again and nearly got shot by a Texas Ranger
orignal
yes, 74 has a bug
orignal
don't use camke
orignal
use make only
orignal
dr|z3d no we use std::atomic
orignal
again I don't support cmake
uop23ip
installed libboost-atomic-dev
uop23ip
Found Boost: /usr/lib/aarch64-linux-gnu/cmake/Boost-1.74.0/BoostConfig.cmake (found version "1.74.0") found components: system filesystem program_options atomic
uop23ip
same error
orignal
stop it
uop23ip
how do i build static without cmake?
orignal
if you want to use cmake
orignal
make USE_STATIC=Y
orignal
if you want to use cmake remove C++20 detection fron there
orignal
make USE_STATIC=yes DEBUG=N
orignal
that's how build
dr|z3d
my bad, orignal
dr|z3d
been a while since I last built i2pd.
orignal
boost 1.74 has a bug with C++20
orignal
and it's in some debian
orignal
dr|z3d btw why you didnt built i2pd recently?
uop23ip
the wiki talks a lot about cmake and so i used it as it worked before i2pd.readthedocs.io/en/latest/devs/building/unix
orignal
<orignal> boost 1.74 has a bug with C++20
orignal
remove 20 from there
orignal
if you need cmake
orignal
in CMakeLists.txt
dr|z3d
lost interest, orignal, after my UI patch bit-rotted.
orignal
CHECK_CXX_COMPILER_FLAG("-std=c++20" CXX20_SUPPORTED)
orignal
remove this line
orignal
bit rotted?
orignal
what does it mean?
dr|z3d
it's like what happens to your feet if you don't change your socks for weeks. only different.
orignal
what was your UI patch about?
RN
*** opens a window to air out the foot stink ***
dr|z3d
it was about 2 months work iirc.
orignal
I don't know what you are talking about
cumlord
*closes a window to marinate in foot stink*
orignal
but it's 2 years ago
dr|z3d
aka bit rot.
dr|z3d
that's about when I lost interest.
orignal
and didn't ask me
dr|z3d
didn't ask you what?
orignal
about that issue
orignal
seems only R4SAS was involved
dr|z3d
we had several discussions that went nowhere.
dr|z3d
you and me.
dr|z3d
"I don't care about UI" I think was where we ended the conversation.
orignal
I don't remeber
orignal
and why did you lose interest?
orignal
to code logic
orignal
i2pd is not anoither UI for Java ))
dr|z3d
I lost interest because it seemed to be a waste of time investing my limited time in something that went nowhere.
orignal
do core stuff
dr|z3d
I do, in Java. that's plenty.
orignal
and UI is really not my priority
dr|z3d
you only needed to build it at the time and take a look at what was on offer. you never bothered. oh well, we've moved on.
orignal
I was not aware
orignal
uop23ip so?
dr|z3d
nevermind, plaz liked it :)
dr|z3d
his words were something along the lines of "this is what I've been missing"
dr|z3d
*** chuckles. ***
dr|z3d
not sure plaz's approval is necessarily a persuasive argument from your perspective, however :)
not_bob_afk
You both have my approval ;)
orignal
not_bob_afk question for you
not_bob_afk
Ask
orignal
how do you handle situation when addresses gets changed at reg.i2p?
orignal
e.g. you receive a new addresses
not_bob_afk
That is a serious problem. I handle it manualy.
orignal
let me explain
not_bob_afk
Why do they change?
orignal
we have a policy if an address is not online for a months it can be replaced
orignal
probably shmoogle.i2p and xmpp.ilita.i2p will be replaced soon
not_bob_afk
Yes. I manually check reg.i2p for new sites manually, and if I see a new site that is "not new", then I replace it manually.
orignal
why do they change? because people abandon them
orignal
why can't you replace autmatically to new one?
not_bob_afk
Java I2P does not allow a way to remove sites via the command line.
orignal
but reg.i2p does and inr.i2p did
not_bob_afk
Yes, they do.
orignal
e.g. address can expire if not online for long time
orignal
inr had 90 days
not_bob_afk
And, yes. They can.
orignal
reg has 30 days ))
not_bob_afk
There is no way to programaticly remove addresses with Java I2P.
not_bob_afk
I have to go into the web interface and do it manually.
not_bob_afk
It's a pain, but doens't happen all that often.
not_bob_afk
eyedeekay: Unless there is a solition to this via the command line.
orignal
I thought you use a spearate database for notbob.i2p
not_bob_afk
But, there has also been talks about how long to set for auto-expire. It was brought up in a meeting a while back.
not_bob_afk
Yes and no. I use the built in router addressbook as that's needed for testing sites.
orignal
it must be base of registrator's policy
not_bob_afk
But, I keep a local database as well for easier processing.
not_bob_afk
But, it is controlled by the router addressbook.
orignal
doesn't it support hosts.txt format?
not_bob_afk
It does.
not_bob_afk
That's what the output is.
orignal
yes update that file from your code
uop23ip
is building, but not finished yet.
not_bob_afk
Yes, the main issue is that there is no good way to interact with the router addressbook.
not_bob_afk
It would be much simpler if it was all text files vs the binary blob that's currently used.
orignal
if it stores in hosts.txt
orignal
that's why I'm asking
orignal
is there a way to use hosts.txt format instead?
not_bob_afk
No, Java I2P does not store it like that. They use a binary blob.
orignal
but you can import from hosts.txt
orignal
delete binary blob and import again
not_bob_afk
Right, yes. But that can't delete sites.
not_bob_afk
I mean, I could?
not_bob_afk
I've never tried that.
orignal
it's just a file
dr|z3d
deleting the db will initialize your hosts, after a restart.
not_bob_afk
Right
not_bob_afk
That would allow me to remove some cruft, yes.
not_bob_afk
But, it would be better if I had a way to remove single hosts.
not_bob_afk
It would be a much cleaner approach.
not_bob_afk
I generally support sunsets on domain names.
not_bob_afk
Though, there has been talk about how that's also a problem.
not_bob_afk
*** shakes his head. ***
orignal
dr|z3d why don't add such functiuon?
not_bob_afk
It's a mainline feature request that I've asked eyedeekay for in the past. As far as I know, he has not added it.
orignal
dr|z3d can implement it in i2p+
orignal
basically one call from UI
dr|z3d
already supported in the UI.
not_bob_afk
dr|z3d: Yes, but how do I do that via the command line?
dr|z3d
but thanks for volunteering me, orignal. I volunteer you take a look at my patch and un-bitrot it. :)
dr|z3d
not_bob_afk: you don't. we've already established that.
not_bob_afk
dr|z3d: Exactly.
dr|z3d
it's possible you could script a form submission, however. not out of the realms of possibility.
not_bob_afk
Anyway, my general policy is that domains don't expire. But, they can change from time to time.
orignal
dr|z3d you patch must be recreated because it's too old
not_bob_afk
dr|z3d: I've tried. It's a pain due to the way the page works.
orignal
so many changes since
dr|z3d
orignal: that's the un-bitrot part.
orignal
so if UI is just a web page
orignal
what's a problem to call it from curl?
dr|z3d
> it's possible you could script a form submission, however. not out of the realms of possibility.
not_bob_afk
orignal: Because it uses features that make that very difficult.
orignal
let dr|z3d make it through simple GET request
not_bob_afk
There is a tag, a "noice"? That must be present in the request to do things. So, I have to pull the page, extract that and then run my request. A pain.
dr|z3d
"nonce"
not_bob_afk
Yeah, that.
not_bob_afk
i2psnark uses them as well.
cumlord
i could probably throw that together
not_bob_afk
I understand why they are there, but it's annoying to deal with.
dr|z3d
mareki2p, meet zzz. zzz, meet mareki2p. :)
mareki2p
dr|z3d, mareki2pb: receiving, over? Yes, I'm here.
dr|z3d
zzz wanted to discuss your snark patch I think./
not_bob_afk
The other issue with a 30 day auto-expire is that some sites do come back after long periods of downtime.
mareki2p
OK, please leave comments in GitLab issue. I think it is better place to discuss that over there. What do you think?
dr|z3d
I think he was waiting for some feedback there, but he'll confirm or deny.
not_bob_afk
This site is a good example. Almost three months of downtime, and then it's back up.
mareki2p
OK, but I'm not sure I notice the comment if it happesn over here.
dr|z3d
mareki2p: I mean he was waiting for some feedback on gitlab, apparently.
dr|z3d
feedback or clarification.
not_bob_afk
Even inr had roughly a month of downtime about six months ago.
mareki2p
Yes, he posted question for me. I answered. I'm monitoring both the GitLab issue and merge request roughly 1x per day or 1x per two days. No updates since then.
not_bob_afk
And then there are sites like stats.i2p and zzz.i2p. If they auto-expire, that allows someone else to take control of the names.
not_bob_afk
*shakes head*
not_bob_afk
Another with about a month of downtime.
not_bob_afk
And another.
not_bob_afk
So, while I generally support sunsets, 30 days is too short. And even with a longer timeframe, some sites should never expire.
cumlord
think i can recycle some parts from multisnark tool if it'd be useful to have delete function with cli, i wanted to make a multi router tool at some point anyway
not_bob_afk
cumlord: That would be very useful.
dr|z3d
sounds good, cumlord
not_bob_afk
Part of my overall goal with my site is to collect historical data on sites. This means keeping old data around forever.
not_bob_afk
But, there are times when it makes sense to not do that.
cumlord
it's on the todo list then :) want to get a little farther on the tracker project then do some copy pasting
mareki2p
Can somebody give me chat permission to ip-dev please? I would like to respond to zzz with "Hi zzz, yes the bugfix is minor, nothing serious. You asked me question and I replied with 3 examples.".
mareki2p
Or copy & paste this message to that chat please.
dr|z3d
zzz's here, he'll see your reply, mareki2p
orignal
not_bob_afk they come back
orignal
but nobody guarantees that address will not be given to someone else
not_bob_afk
Ahh!
not_bob_afk
So, you continue to monitor them, but allow others to take the name if it's been dead for too long.
not_bob_afk
Understood.
orignal
we have some reserved addresses
not_bob_afk
I need to take leave. I'll be back here later in a few hours.
orignal
and can't be taken
not_bob_afk
Yes, that's the way to deal with it.
orignal
the problem is schmoogle
not_bob_afk
But, I still think 30 days may be too short.
not_bob_afk
Oh?
orignal
that guys wanted to show fuck to everybody
orignal
we are going to reaplace it to new one
orignal
*guy
not_bob_afk
I see that site has been dead for roughly two weeks now.
orignal
yes, two more weeks and will be repalced
orignal
ofc we can reaplce it even now
not_bob_afk
Perfect.
orignal
but don't to vilate own rules
not_bob_afk
Do you have the new address for it now?
orignal
not yer
not_bob_afk
Ok
orignal
will build later
not_bob_afk
Message me when it has a new address and I'll update.
orignal
same with xmpp.ilita.i2p
orignal
will do
dr|z3d
why is shmoogle a problem?
not_bob_afk
Good to know on that.
not_bob_afk
Thank you.
not_bob_afk
*** is really AFK now. ***
orignal
dr|z3d because we want to keep it
dr|z3d
if you're planning on a new search engine, you can think of a better name than shmoogle.
orignal
not a problem to start an instance of YaCy
orignal
shmoogle is well known now
dr|z3d
it's well known for being abandoned.
uop23ip
running well so far, thanks orignal
uop23ip
notice for myself: do not mention the "cmake" ;)
orignal
wrong
orignal
it need to fixed if boost if <= 1.74 don't use c++20