IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#i2p-dev
/2025/04/26
@eyedeekay
&zzz
+R4SAS
+RN
+StormyCloud
+T3s|4
+acetone
+dr|z3d
+eche|off
+hagen
+not_bob_afk
+orignal
+postman
+qend-irc2p
+segfault
+snex
+wodencafe
Arch
BubbRubb
Danny
DeltaOreo
FreeB
FreefallHeavens
Irc2PGuest13713
Irc2PGuest29468
Irc2PGuest373
Irc2PGuest58507
Irc2PGuest59134
Irc2PGuest8291
Irc2PGuest86267
Onn4l7h
Onn4|7h
Sisyphus
Sleepy
T3s|4__
ardu
b3t4f4c3___
bak83_
boonst
cumlord
dr4wd3_
eyedeekay_bnc
mareki2p_
poriori_
profetikla
rapidash
shiver_
solidx66
theoddone
thetia
u5657_
uop23ip
w8rabbit
weko_
x74a6
orignal I will check
zzz eyedeekay, do you plan to migrate the old MRs or not? I need to start doing new ones but I don't want them buried under the old ones
zzz altonen, was looking at your proxy, the Host vs host header thing is weird, seems like a java-side bug?
altonen yes, zzz.i2p and i2pforum.i2p wouldn't load with lowercase header names, had to convert them to header case
zzz hmm we should be canonicalizing them when processing them in the server proxy. I'll investigate
zzz however, you should be sending the b32 in that header, to not expose your local addressbook name to the server
zzz the server side proxy swaps it out with its local hostname to make the webserver happy
altonen ok i'll fix that, thanks
zzz rust making them all lower case is a little bizarre though
altonen it is, i agree but they say it's spec-compliant and easier to handle in code AFAIR
zzz sure, can't really blame them, just not typical
zzz ok, I see it, we weren't canonicalizing the host header, I'll fix it
zzz so your host header was going through and we added a Host header so the server got both and complained
altonen that explains it, it was an intereesting bug for sure
zzz another thing for your list, doesn't look like you're saving/persisting and returning the last-modified/etag headers in subequent requests?
altonen i think i am because at some point i stripped them and then brought them back
zzz please point me where and I'll take a peek
altonen proxy/http/mod.rs has ILLEGAL vector with the headers that are stripped, rest are kept as-is
zzz yeah that's in the proxy. But for the subscription fetches, are you saving the last-mod/etag in the response (both in-mem and on disk), and sending them as if-mod-since/if-none-match the next time, so you don't get the whole hosts.txt every time?
altonen i'm not
zzz ok, that's for your list then )) stats.i2p will only send you the delta from what you had, if you include the header, so it's important for subscription site bandwidth
altonen will add, ty
zzz low priority unless you get popular ))
orignal updated reseeds
orignal zzz, what's with memcpy?
zzz still havent heard from hottuna, still serving RIs from Jan. 1, afaik
orignal so, do we exclude it?
orignal becuase it was not mentioned in your e-mail
zzz yes. pretty sure we already had this discussion weeks ago. Forgot to put it in the email
orignal thanks
orignal will remove
zzz altonen, you should exclude it also. Everybody should be checking timestamps in the version of received reseed su3's
zzz reseeds get 'stuck' where the reseed server is up but the router went down so they never get new RIs
orignal that's what we do ))
zzz check both the su3 version which is a timestamp, and the timestamp of each RI, throw out things too old
orignal I do it if more than half
zzz yeah but if the su3 version is old you can just throw the whole thing out before you check each RI
orignal will add this check
zzz just checked, memcpy is now completely down
zzz the version for reseed su3's is full unix timestamp (ms) in ascii
zzz wait, sorry, in seconds
orignal don't worry I will check what we do it reseeder
orignal since it seems I'm in charge over it now
orignal *** is a pythin developer )) ***
zzz I finally heard back from diva.exchange, he's going to merge in your changes into his fork, he's the last one that hasn't added the version and caps checks
orignal if mine and R4SAS's reseeds works properly
zzz sure they do
zzz diva reported a big reseed spike/attack yesterday, peak of 3200/hour
zzz did you see anything?
orignal mine is ygg
orignal need to ask R4SAS
zzz his normal rate is about 500/hour
zzz eyedeekay, did you see any spike yesterday on yours?
zzz correction, diva has a fork of the go reseed, not the python one
orignal even better
eyedeekay I saw a bunch of weird stuff yesterday including a spike in reseeds, but I'm seeing wild amounts of traffic on every clearnet service and most of my I2P services now
eyedeekay Pretty sure bots
eyedeekay zzz sorry I thought we had discussed this already, go ahead and create MR's for everything you need
eyedeekay I will be migrating only the closed MRs
eyedeekay That way we have records of all the MR's we've merged or rejected but nothing new gets buried
zzz ok so to double-check, you're not migrating open MRs, I should open new PRs based on the the branches in my account which did get migrated
zzz right?
eyedeekay Yes exactly
eyedeekay There should be no interference from the migration of old MRs that way
zzz ok. we had certainly discussed it but had not heard your decision
zzz so what will be incoming is:
zzz 2 old WIP PRs
zzz 1 old PR for 2.9
zzz 2 new PRs for 2.9
zzz several new PRs either WIP or for 2.10
eyedeekay All right sounds good, I will be watching for the new PRs.
eyedeekay Of reseeds: I am going to check and try to see if there is a difference between how much they're hitting the html pages vs how much they're hitting the su3, I wonder, are the other reseeds seeing attempts to scrape things they aren't necessarily serving...
zzz ok. see also email I cc'ed you on
zzz can you fix the gitea CI enough that the build runs and the others don't, so we stop getting red x's?
eyedeekay Yes I will
zzz thanks. Also please holler when you have a custom homepage, I'm awaiting that before tweeting
eyedeekay Ok shouldn't be much longer on that, probably will drop it in and deploy it this evening
zzz ok. I got the 3 old MRs re-entered from the existing branches. Will start working on the new ones
zzz eyedeekay, it won't let me push a new branch to my account to prep for a new PR. Do I have to branch via the website? Can't figure out how if so. Or do you need to allow me to create new branches by pushing?
zzz ok, branches -> create new branch from master
eyedeekay Hm that seems wildly off, neither should be the case, you should be able to just push branches
eyedeekay I will figure out how to make sure you can do that
zzz maybe it was just a transient fail
zzz stand by
zzz yup. sorry, false alarm. new branch push works
dr|z3d eyedeekay: use nginx's throttler to kill excess requests.
dr|z3d you could hook that into fail2ban so you snare the bots.
dr|z3d another method is to check that any requests are sending a cookie header, and block if they're not, assuming the bots aren't sending those.
eyedeekay I've actually got pretty tight limits on the su3 itself
dr|z3d I meant for gitea, mostly, not so much the reseed server.
eyedeekay oh yeah that's on the menu soon
dr|z3d (assuming that's where the bots are running amok)
eyedeekay One of many
eyedeekay They like the mirrors alot too
zzz ok, 4 new branches and PRs done, now 7 PRs total, 4 WIP, 3 for this release
dr|z3d gitea's down right now, inentional, eyedeekay?
eyedeekay Yeah will be very temporary