+R4SAS
+RN_
+mareki2p
+orignal
+postman
+qend-irc2p
Arch
Birdy
Daddy_I2P
Danny
FreefallHeavens
Irc2PGuest95549
Irc2PGuest96214
NiceBoat
Onn4l7h
Onn4|7h
Over
Romster_
Sisyphus_
Sleepy
StormyCloud
Zapek
_bob_
aargh3
acetone_
ahiru_
anontor
b3t4f4c3___
cims
dfhj
dr4wd3
duanin2
eche|off
eyedeekay_
f00b4r_
hagen_
makoto
marek
marek22k
mesh
n2_
nilbog
not_bob_afk
nyaa2pguy
o3d3
poriori
profetikla
r00tobo_BNC
rapidash
rednode
solidx66
sublimia
test3847473
uop23ip
urist_
vivid_reader56
wodencafe
x74a6
zelgomer
zzz
obscuratus
eyedeekay: PeerSelector is a more involved issue than I first appreciated.
obscuratus
I was reviewing the FloodFillPeerSelector class, and I couldn't see anywhere where it understood how to use any other netDb except for the primary floodfill netdb.
obscuratus
On the plus side, things seem to work OK dispite this being somewhat borked.
obscuratus
We may need to pivot to something like relying on trip-wires for RI coming in the Inbound Message Distributor.
obscuratus
As a work-around, on my testing network, it wouldn't be difficult at all to make sure the subDbs are populated exhaustively with every FF RI.
obscuratus
Or, maybe begin testing the nested netdbs without any RI at all, and make sure nothing breaks when we run that way.
obscuratus
I'm almost running that way now, with only the minimum 3 FF in each subdb.
eyedeekay
Thanks for the update, my next move will probably be in the direction of no RI's in subDb's first