IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#ls2
/2022/03/12
eyedeekay So this is pretty uncool:
eyedeekay I think we should try to appeal to stewards-p2p@wikimedia.org on behalf of our users, it looks like they're banning every IP address that they can enumerate, or whoever is providing them with their list
eyedeekay So I'm going to draft an email
eyedeekay It looks like if you're not firewalled, you end up on this list
zzz orignal, R4SAS, the crash problem was that it trying to decode the "i" value in SSU? if we change it to "k" would that prevent the crashes?
R4SAS zzz: hm, I dont remember
R4SAS if I correctly understand, SSU2 value was unexpected
R4SAS zzz: ^
zzz right, so if we change it to "k" will that prevent the crash?
orignal zzz, older version assumed that i can appear for NTCP2 only
orignal no, leave as "i"
orignal it's fine
orignal not only "i" but also "s"
zzz can we change them to prevent the crashes?
orignal why should we?
orignal 2.41 is fine
orignal nobody cares about older versions
orignal and alomost everybody has upgraded
orignal I have another question
orignal does you SSU2 support ipv6 now?
zzz by my measurements only about 45-60% of i2pd routers have updated
zzz re: ipv6, all my testnet testing is on ipv6. I actually have only tested ipv4 in standalone test code
orignal I mean one in main network
orignal then let thm crash
orignal or ypu have another option
orignal publish SSU2 rather than SSU
zzz I can't do SSU2, unfortunately.
zzz so we can either rename i and s, or let them crash
zzz you want to crash them, ok
orignal let them crash then
zzz ok, probably later today
orignal and I will start publishing SSU2
orignal probably with caps="6"
orignal for the beginning
orignal or if you run it as non-FF it affects less
zzz right, disabled ff a week ago
zlatinb eyedeekay: my home routers are firewalled and I can still browse wikipedia
zlatinb pls collect as much evidence as you can, we can get TF to write about this if it's real
Xeha the complaint was about being a editor on wikimedia projects
zlatinb oh, well then I can't check
Xeha i checked a few of mines where i2pd runs, all negative (ie not blocked)
zlatinb are they firewalled or OK ?
Xeha of course they arent firewalled :) thats the point of FFs
zzz zlatinb, I've been testing with the reduced tunnel build expiration for a week, no issues
zzz where do you want to go from here?
zlatinb what values did you test with?
zzz + static final int REQUEST_TIMEOUT = SystemVersion.isSlow() ? 10*1000 : 5*1000;
zzz + private static final int LOOP_TIME = SystemVersion.isSlow() ? 1000 : 800;
zlatinb 800? I thought you wanted 500
zzz not iirc
zlatinb I'm happy for this to be committed. I won't be able to test with MuWire until it goes into Maven but I can test initial startup times locally
zzz ok, I'll push it
zlatinb and as a separate exercise we'll have to ask echelon to install linux on his M1 so we can distinguish that as not slow
zzz isSlow() will still return true until we have jbigi for it
zlatinb we have jbigi for aarch64
zlatinb linux-aarch64*
zzz oh, linux, yeah
zlatinb macos-aarch64 isn't a thing yet unless the user builds manually
zzz just need somebody to do it
zlatinb I was looking at buying an M1 yesterday but decided to wait at least until vmware comes with a hypervisor for it
zlatinb I don't like the idea of dual-booting
zlatinb the mac mini is the cheapest option for getting an M1 even if its usability is limited
eyedeekay I'm working on figuring out what it is, so far about half of the routers I've tested have been in the list for one reason or another
eyedeekay Of my routers 4/5 that I've checked have been in the blocklist
eyedeekay I'm going to test against a big sample of router IP addresses and try and determine statistically if we need to be very concerned
eyedeekay so far, Of the 304 IP addresses I tested, 136 were banned
zzz at least 3 SSU2 routers on live net, no issues so far
eyedeekay Still asking a few others to check if their peers are banned from Wikipedia but I'm pretty convinced their "API Partners" are recommending they ban I2P routers and providing them with a way to query a pre-enumerated list
eyedeekay I think the list will turn out to be of limited reliability, but possibly in a consistent way.
eyedeekay If I am right one point of evidence should be that the fraction of known peers who are banned should be about the same
R4SAS eyedeekay: who operates i2pmail.org gateway
eyedeekay postman, I don't know who postman is but that's who to reach out to AFAIK
R4SAS ok, I wrote issue on @mail.i2p, let's see if anyone response
R4SAS #mail.i2p ***