@eyedeekay
&kytv
&zzz
+R4SAS
+RN
+StormyCloud
+acetone
+altonen
+dr|z3d
+hagen
+hk
+mareki2p
+orignal
+postman
+radakayot
+segfault
+snex
+weko
+wodencafe
Arch
Danny
DeltaOreo
FreeB
FreefallHeavens_
Irc2PGuest12011
Irc2PGuest12735
Irc2PGuest17807
Irc2PGuest18076
Irc2PGuest59134
Onn4l7h
Onn4|7h
Sisyphus
Sleepy
T3s|4_
T3s|4__
Teeed
aeiou
ardu
b3t4f4c3__
boonst
cumlord
death
dr4wd3_
eyedeekay_bnc
not_bob_afk
onon_
phil
phobos
pisslord
poriori
profetikla
qend-irc2p
rapidash
shiver_
solidx66
thetia
u5657
uop23ip
w8rabbit
x74a6
zzz
those two IPs are definitely banhammer candidates, they are persistent
dr|z3d
so maybe a counter and incremental ban times?
dr|z3d
or you're just thinking add them to the blocklist and be done with it?
zzz
the change today catches it but perhaps a newsfeed hammer is appropriate since the release is a month out
dr|z3d
yeah, I've just pulled in that commit, I think I might increase the ban period, and some explictly logging there to indicate the temp ban is probably a good idea.
dr|z3d
*explicit
zzz
eventpumper warn logging you will see them
dr|z3d
74.164.x.x on your offender radar?
dr|z3d
it's not that I can't see the blocks, more the question that it's not explicitly stated in the logs that the offending ip is being temp banned.
orignal
zzz do you drop LeaseSet with cryptotype=1 now?
orignal
in address I mean
zzz
WARN [NTCP Pumper ] ter.transport.ntcp.EventPumper: Blocking accept of IP with count 3: 110.137.220.12
zzz
^^ temp ban
zzz
it's interesting that two of them spun up at once, one AWS canada and one indonesia
zzz
we did implement the full 'probing resistance' recommendations from the NTCP2 spec, that helps
orignal
because I see ton of "Destination: Couldn't find published LeaseSet for "
zzz
orignal, LS with type 1 keys?
orignal
yes, type 1 in crypto key type in cert
zzz
oh in the cert. looking...
orignal
seems you have changed something recently
orignal
because it worked fine for years beofre
dr|z3d
>> WARN [ NTCP Pumper] …ntcp.EventPumper: Blocking NTCP connection attempt from: 185.207.242.63 (Count: 2)
dr|z3d
If we're blocking for 43 minutes, then we probably only want to see the logged event the first time.
zzz
not debating log policies, do what you want
dr|z3d
sure, sure, just floating the idea, no debate required :)
zzz
:)
orignal
I just don't want to change the address
zzz
I don't think I changed anything orignal, but give me a b32 and I'll try it
orignal
jw4533ydxwgmna6wrvw6n55h3aem3vrqilhxjkmpgrmjfhldlscq
orignal
it fails on publishing
zzz
I can't get the LS
zzz
but FYI, most ffs out there are i2pd since a year ago, due to a bug in the auto-ff-enable code where we required SSU 1 to enable !!!
zzz
we will have lots more ffs after the release
orignal
so you think that i2pd drops such LSes?
orignal
interesting
zzz
don't know
orignal
but it worked fine until couple weeks ago
zzz
try force-storing it to a ff you control, if you have a way to do that
orignal
will check
orignal
maybe it's my bug
zzz
we might double or triple the ffs after the release. It's a really dumb bug on my side
zzz
as for enc types and cert handling, I have a ton of changes locally for PQ but none of that is checked in
orignal
will inverstigate anyway
orignal
because it might be related to my PQ changes
orignal
forgot about that crap with crypto type 1
orignal
// Java floodfill never sends confirmation back for unknown crypto type
orignal
I see this in comment
orignal
is it true?
orignal
looks like I'm trying to publish on a FF that's not online or something
zzz
will research and let you know
orignal
so, cam we assume there are only 2 valid values for destinatiion in this field 0 and 255?
zzz
orignal, I have a type 1 dest from you in 2020 in my addressbook, I also created a new one and added it, and I found a code comment that we allow types 1-3
zzz
so the basic parsing still works. I can't easily verify that a LS store will work
orignal
so, Java does send reply to 1. right?
orignal
also publishing failure for type 1 is my bug
zzz
I can't test it for sure, but I don't see any problems or recent changes
orignal
no it was my recent change about PQ
orignal
however that code without confirmation is old one
orignal
I should remove it
orignal
jw4533ydxwgmna6wrvw6n55h3aem3vrqilhxjkmpgrmjfhldlscq works fine now and no issue with publishing
zzz
got the LS
zzz
we report: Unsupported encryption options
zzz
remote: fatal: unable to write loose object file: No space left on device
zzz
error: remote unpack failed: unpack-objects abnormal exit
zzz
^^ eyedeekay StormyCloud
StormyCloud
Hmm
StormyCloud
Am I being summoned
zzz
so we need to upgrade to the stormy-large instance? :)
StormyCloud
Ah space, eyedeekay let me know if it’s a space issue or configuration thing
orignal
but do you add such LeaseSets into netdb even with unknown encryption option?
zzz
yes
orignal
good
eyedeekay
Ah it's a config issue, I was keeping cached data for too long, should not be long to fix
eyedeekay
forum.gitea.com/t/how-to-configure-cron-task-for-delete-all-repositories-archives-zip-tar-gz-etc/4848/2
eyedeekay
OK, changes made to prevent that one happening again, I'm going to review the rest of the schedulable maintenance today and make sure it's all configured to run frequently enough
eyedeekay
Postmortem: Found that gitea keeps copies of every repo "archive" it's generated until a cron job runs to delete them. It generates an archive in advance of a person needing them at what appears to be checkin time. This results in hundreds of zips and tar.gz copies of the code laying around in /var/lib/gitea/data/repo-archives. The cron job did not run often enough to delete the old ones before the disk
eyedeekay
exploded. Increased frequency of all cleanup cron jobs.
eyedeekay
Problem was likely exacerbated by mass-mirroring as part of the migration process, tons of checkins on tons of repos at once
eyedeekay
Now back to fixing the base CI container so apt knows where the packages are :/
zzz
thanks eyedeekay
eyedeekay
No problem, sorry about the breakdown
zzz
minor blip
dr|z3d
yeah, watch out for those archives, it quickly suck your storage dry :)
dr|z3d
*can
eyedeekay
Just glad to find that the config options I needed were already there
RN
Irc2PGuest47673, tell us about your snark issue: OS, install method, router version, etc
Irc2PGuest47673
Hi. I just noticed a small problem in i2pSnark. Torrents are divided into several pages. If you are on any page with inactive torrents, the bottom bar will display the current download speed (correctly) but will not show the upload speed (incorrectly). When you switch to a page with at least one active torrent, both speeds are displayed correctly. I am using GNU/Linux with version
Irc2PGuest47673
i2psnark-standalone 2.8.0 from the skank.i2p eepsite
RN
dr|z3d, this one's for you (at least at start)
zzz
thanks for the report Irc2PGuest47673