IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#saltr
/2022/06/28
mesh term99: what is eephead?
mesh oh I see
mesh well that's werid.... but makes sense
mesh basically load the page every hour or so. but it probably makes sense just to set the backup tunnel count to 1
mesh right now I have backup tunnel set to 0 to 'save resources' but when I do this it takes quite a while for the site to come alive
mesh ~5-10 minutes
mesh I wonder if I had some backup tunnels that would remove this problem
dr|z3d ~/i2p/eephead address.i2p
dr|z3d backup tunnels don't do much of use, they're just additional tunnels. set tunnel allocation to 4 in/out, drop to 2 after a period of inactivity. should be fine.
dr|z3d if the service you're accessing is remote and you want to keep it responsive, as term suggests, just eephead the service via crontab every 10 minutes to ensure you have the leaseset.
dr|z3d so, in summary, the issue is with the client acquiring the leaseset for the service more than anything else, not the "liveliness" of the tunnels per se.
dr|z3d this all presupposes you've installed xubuntu of course, mesh. you can do it on windows with more effort.
parabo dr|z3d, Ellenor itsjustme
parabo hey dr|z3d. I have a beard
Ellenor cool, now go blow up a russian orthodox church
parabo that's terrorism
parabo hi mik
parabo Ellenor, how are you?
Ellenor gluten poisoned
parabo have you tried pantoprazole?
parabo hi mik
mesh dr|z3d: hmm, that kinda sucks
mesh dr|z3d: you're saying the problem is on the client... not the server. clients that don't use a service for a while will end up in this state unless they eephead the service every 10 minutes?
mesh there's gotta be a better way... it seems to me after a few days, I end up in this bad state where my site refuses to load for a good 5-10mins
mesh and this is on a very popular router participating in thousands of tunnels. you would think one benefit of running such a flood fill router is that my sites would load fast :)
mesh maybe there should be a way for floodfills to prioritize distribution of their own local destinations to the other floodfills
mesh I think that would be a good "reward" for running a floodfill
dr|z3d iirc we already do prioritize distribution of leasesets for our own services, though I'll have another look when I have a free moment.
mesh I was planning to write some code that would periodically do a health check on some services... that might help.
mesh man I wish I wasn't so sick. I wonder a LeaseSet is just an Advertisement right... can they be cached?
mesh oh yeah, of course, a LeaseSet isn't an Advertisement for a Router... it's an Advertisement of a Tunnel and expires as soon as the tunnel expires.
dr|z3d it's a collection of leases with a lifespan of approx 11 minutes, if you include the grace period.
dr|z3d 1 lease per tunnel.
mesh I guess it really is just a matter of polling the i2p destination, constantly trying to build tunnels to it, to keep a viable LeaseSet around.
mesh it's going to lead to a very poor user experience. Unless perhaps would it help to spam out a bunch of HEAD requests concurrently with the first GET request...
mesh man why is Apache Tika so complicated
dr|z3d ok mesh, latest dev build pushes out leasesets to more floodfills, see if that helps your issue. does more in parallel too.
dr|z3d it was already more than vanilla, now it's more++
dr|z3d we also now spend a bit more time looking for routerinfos and leasesets, maybe that will also help.
deaddrop <dr|z3d> I was raping my nephew yesterday and i totally came in his mouth lmao
march <dr|z3d> I want to rape a 4yo boy the most lmao