zzz
orignal, it's fixed so you can't create thousands of tunnels anymore with i2pd?
dr|z3d
oh, sim608.
dr|z3d
if he's on irc, great, hopefully he takes the hint.
orignal
no, BOB is fixed
orignal
to replace previous session rather than create new one
zzz
what, so now it's even better at creating thosands of tunnels? (((
zzz
oh ok
orignal
you can create thousands of tunnels
orignal
if you need to
dr|z3d
"After some time, on a neighboring computer included in the same router as Linux with I2PD, the Internet begins to fall off slowly until the router reboots."
zzz
can you put a limit on it to stop the idiots?
orignal
I don't see your point
zzz
to stop terrible bob/sam applications from clogging up the network
orignal
how much?
zzz
or you think this is just an i2pd bug, not a client-side crazy application?
orignal
I can a limit parameter to this config
orignal
this is reroshare bug
zzz
Blinded message
zzz
Blinded message
orignal
they did it wrong
orignal
and i2pd also handled this situation worng
zzz
so we default limit to 100 destinations, with an router option to override
orignal
will do
zzz
won't prevent an intentional attack, but might prevent something stupid
zzz
thanks
orignal
will set to 500
dr|z3d
I'm a bit more relaxed than that, but not much, on the defaults.
dr|z3d
256 I've got here.
orignal
I'm pretty sure people use more than 100 already
zzz
for what?
dr|z3d
notbob's scanner will hit several hundred quite easily.
orignal
to shit on kislitsa ))
orignal
from different addresses
zzz
lol
orignal
I know because they write about it
zzz
I added our limit in 2014, and nobody has ever complained, and I don't think the option is even documented
orignal
but tell me why number of tunnels is a problem?
dr|z3d
you hit the limit, it gets logged as a warning iirc.
dr|z3d
with the config option.
zzz
it's doubling the number of tunnels and number of leasesets in the network.
zzz
which doesn't seem to be a problem for now, but if he gets a bunch of his friends to do it also, it might be a problem
dr|z3d
only needs half dozen cretins running broken retroshares for bad things to start happening.
zzz
it's also wasted a huge amount of our time trying to figure out whats going on
dr|z3d
I think I'm going to bump max session to 512. enough to accommodate pretty much all legitimate uses without allowing things to get out of hand without manual configuration.
dr|z3d
does a single server running on a router count as a single session?
dr|z3d
say a webserver, for example, not necessarily hosted on jetty.
dr|z3d
ah, you asked him on the forum, zzz. probably better to nudge him to update his retroshare for better performance.. more persuasive :)
orignal
there is also issue on github
dr|z3d
"if you don't want to see your router crash repeatedly, updating retroshare might be a good idea"
zzz
I don't know enough to give any other nudges
dr|z3d
well, his router's crashing, orignal's suggesting he's using a buggy retroshare + bob combo.. that's enough. if a newer retrsoshare fixes the issue, everyone wins and he feels like he's won something. just telling him to stop doesn't incentivize him.
zzz
join the party over on 333.i2p or GH then. I don't know enough to provide nuanced advice
orignal
my suggestin is again
dr|z3d
also, chceck orignal's bug report.
orignal
use SOCK+inbound tunnel
orignal
like we did from day one
orignal
and it works
dr|z3d
it's amusing.. "After a day of work, virtual memory is more than 20 G."
orignal
"virtual memory"
orignal
he is just a noob
orignal
zzz what language do you see at 333?
orignal
I remeber if it was a traslation or I replaced texts directly
zzz
ru. we block accept-language header in the proxy
orignal
yes
orignal
that's why I'm asking ))
orignal
333 is modified eldorado
orignal
the translation is mine ))
orignal
but being drunk if I remeber right
zzz
heh
orignal
but pretty identical to yours
orignal
btw maybe you can advise
orignal
intiially I have made a strategic mistake by using sqlite for database there
orignal
now want migrate to mysql but don't know how
orignal
I though eldorado can do it somehow
orignal
it's not just database also update configs of eldorado etc
orignal
maybe drivers
orignal
like ruby gems etc
zzz
I don't remember anything, it was 15 years ago. I'm still on sqlite, db is still only 20 MB
orignal
that's fine
orignal
it's not about size, I just don't now sqlite commands ))
orignal
*know
zzz
definitely don't ask me for advice, I know nothing ))
dr|z3d
well, if you can migrate the content db to mysql, you can probably reinstall eldorado and point it at the converted db. probably worth testing on a dev box.
orignal
me too
orignal
I even don't remeber how I comipled ruby of that version
orignal
dr|z3d the problem that eldorado requires some old version of ruby
dr|z3d
sounds like fun :) maybe run it in a container of some sort, snap or whatever.
orignal
who cares ))
dr|z3d
unlikely, but triz isn't pimptrizkit?
orignal
gents I see bunch of routers like [2602:fc05::24]:7777
orignal
9 so far
orignal
I'm pretty sure the all belong to the same owner
obscuratus
orignal: Aren't those StormyCloud routers?
orignal
maybe, just curious
dr|z3d
if they are, they should have a family attached.
orignal
the question if
orignal
why all IP from the same range
obscuratus
dr|z3d: Yup, that's why I noticed.
orignal
that makes me confused
orignal
same subnet
dr|z3d
the why is because that's the ip range he owns :)
orignal
owns?
orignal
how can you own an ip range?
dr|z3d
if you're only seeing 9, you're not getting the full picture.
orignal
no, I see at least 9
dr|z3d
ips and ranges can be bought and sold like any other commodity :)
obscuratus
It looks like StormyCloud is usually firewalled on IPv6. But a few routers seem to think they're not firewalled. Much more routers on the IPv4 address range.
dr|z3d
I see around 80 routers here.
dr|z3d
orignal's obviously impressed by the vastness of StormyCloud's enterprise, obscuratus ;)
StormyCloud
IPv6 isnt firewalled, but you said there might be an issue with the pv6 peer test
StormyCloud
but yes I "own" an IPv6 and IPv4 range
dr|z3d
single ips are now going for, what, $50 each, StormyCloud?
dr|z3d
of course you can also rent them. just to confuse orignal even more :)
StormyCloud
I have a /22 block I am renting for 0.53 per IP
orignal
I noticed it on ipv6 only router
dr|z3d
orignal, do you throttle floodfill leaseset publication?
orignal
no
dr|z3d
so in i2pd-land, an infinite number of multihomes is possible then.
dr|z3d
except java i2p throttles which enforces a 4-5 max limit (3 hops).
orignal
how can you recognioze multihoming at FF?
dr|z3d
by destination.
dr|z3d
ff sees foo.i2p, there's a limit to the number of leases it'll process before it drops requests to publish in a given timeframe.
dr|z3d
wb zzz
dr|z3d
spike central on the part tunnels happening again.
orignal
number of leases is 16
orignal
max
orignal
but multihoiming is different story
dr|z3d
per period? so you do throttle then.
dr|z3d
or?
orignal
16 leases in one LeaseSet
dr|z3d
sure. that's a fixed limit.
orignal
probably I check if one if get updated to frequitly
orignal
al least a second
dr|z3d
ok, so that's a throttle of sorts, but nothing that will limit a client dest to 4 3-hop multihomes.
dr|z3d
I'm just wondering if having different/no limits on java and i2pd floodfills is detrimental to the network.
orignal
if you tell me how
orignal
I receive a new LS how do I know if it's the same instance or another
dr|z3d
you don't. all you can measure is the frequency it's being published to your floodfill.
orignal
but I still don't know number of instances
dr|z3d
if you work on the assumption that a single instance will publish every 10 minutes, you've got a rough idea of number of instances, though that's not the specific point.
dr|z3d
you're not throttling the number of instances per se, you're throttling the number of publication requests for a given dest.
dr|z3d
or at least java i2p is.
orignal
this assumption is worng
orignal
i2pd publishes as soon it builds new IB
dr|z3d
ok, maybe I've got the details slightly wrong. but the general point is there's a predictable cadence for publishing new leases.
dr|z3d
given each lease lasts 10 minutes, or 11 minutes with grace period, there's a quantifiable number of leases coming from a single host.
dr|z3d
zzz: any merit to ensuring i2pd and i2p are on the same page wrt floodfill lease publication limits? any downside to the current mismatch?
zzz
no
dr|z3d
so it doesn't matter than i2p's throttling leasests puts on floodfills and i2pd isn't? doesn't that make the max number of multihomes a bit of a lucky draw?
zzz
if you're trying to convince i2pd to do something, fine, good luck, I don't have time to join the party
orignal
zzz, huh?
orignal
I suppose if there is a problem I try to resolve it
dr|z3d
I'm trying to work out if synced limits for publication of leasesets to floodfills is of benefit to the network, and what the downsides are with the current ad-hoc arrangement.
orignal
but I can't understand the problem so far
dr|z3d
if having a network wide limit benefits the network, then I'm sure orignal will consider it.
zzz
whatever you're proposing, state your case, I have nothing to offer
dr|z3d
what I'm proposing, if it's not already clear, is an agreed network-wide limit on publication of leases to a floodfill in a given timeframe, with a throttle as i2p already implements, to make the limit of multihomes that can reliably be hosted predictable without any measure of guesswork involved due to the difference between i2p and i2pd.
orignal
I can trottle by time but not number of instances
orignal
because I can't recognize them
dr|z3d
yeah, that's what i2p does, enforces a limit time-wise.
orignal
what's the limit?
orignal
in seconds
dr|z3d
let me dig out the relevant code.. one sec.
dr|z3d
I'll get back to you on that, orignal, it's evading my searches right now.
orignal
no rush
orignal
btw SSU2 produce false "Firewalled" much rarely than SSU1
dr|z3d
Note to self: gitlab.com/i2pplus/I2P.Plus/-/blob/1864f07b928e52d6952906cf8bd426b8fb4a8d22/router/java/src/net/i2p/router/networkdb/kademlia/LookupThrottler.java
dr|z3d
orignal: gitlab.com/i2pplus/I2P.Plus/-/blob/1864f07b928e52d6952906cf8bd426b8fb4a8d22/router/java/src/net/i2p/router/networkdb/kademlia/LookupThrottler.java
dr|z3d
I *think* that's the code.
orignal
50 lookups per ... ?
dr|z3d
though it's for lookups not publication. so hmm.
orignal
per 3 minutes?
dr|z3d
3 minutes.
dr|z3d
(clean time)
dr|z3d
yes
dr|z3d
30 in default java.
dr|z3d
(lookups that is)
orignal
no. 50 lookups per 3 minutes?
dr|z3d
50 in 3 minuutes before the throttle kicks in (i2p+) or 30 lookups / 3 min (vanilla i2p).
orignal
R4SAS
orignal
we must mind it for reg.i2p
orignal
I think it's too small