@RN
@RN_
@StormyCloud
@T3s|4_
@eyedeekay
@not_bob_afk
@orignal
@postman
@zzz
%Irc2PGuest68665
%Liorar
%mareki2p
%snex
+FreefallHeavens
+Onn4l7h
+Onn4|7h
+bpb
+leopold
+poriori
+profetikla
+qend-irc2p
+r00tobo
+uop23ip
+xHarr
Arch
BubbRubb
Danny
DeltaOreo
H20
Irc2PGuest2981
Irc2PGuest90985
Meow
Over1
Stormycloud_
T3s|4
ac9f
acetone_
anontor
dr|z3d
duck
floatyfloatilla
halloy13412
mahlay
makoto
nZDoYBkF
nilbog
ntty
r00tobo[2]
shiver_
simprelay
solidx66
thetia
u5657
vivid_reader56
zer0bitz
dr|z3d
you know you can stat/graph the snark tunnels, and if you're using the prometheus plugin, also have grafana graphs..
cumlord
haven't given that plugin a try yet but forgot to turn stats on that router for it
dr|z3d
stats should probably be on, you just need to toggle graph generation.
dr|z3d
if you like graphs and stuff, definitely recommend the prometheus plugin + grafana.
cumlord
yup i never did for that router or manually had it all turned it all off, can't remember lol
dr|z3d
some graphs are configured by default for basic perfomance stats, some you need to enabled advanced stats mode to get at.
cumlord
damn complete fail on the baby test router though, maybe didn't give that enough ram
cumlord
was fine after start but now has job lag of 24 seconds
dr|z3d
fail? how so?
dr|z3d
if you're hosting a ton of torrents, 512MB is probably the least you can get away with, 1G better. depends on what you're doing.
dr|z3d
give it time to settle down, and review /jobs and /jobqueue to see what's causing the lag.
dr|z3d
(sorting by max column on /jobs will indicate anything taking more time than it should..start there, then check avg).
cumlord
message delay* is very high on that one, causing other tunnels to fail
dr|z3d
how many available cores/threads?
cumlord
i suspect it's the lone external snark running on it, embedded and others are disabled on it
dr|z3d
suspect tells us nothing. investigate job queues, logs.. :)
cumlord
load on it is very high for apparently no reason lol, it's very slow on console atm but coming down after stopping the snark instance
dr|z3d
cores/threads?
cumlord
4 cores, 8 threads, 16 jobs in queue
cumlord
maybe i should manually set jobs to half that or something?
dr|z3d
ok, and max jobrunners configured?
dr|z3d
no, maybe you shouldn't.
dr|z3d
if anything, you want to remove any override you've configured and see how it performs then.
cumlord
it's using the default job runners, i set no override
dr|z3d
handling jobs, is, for the most part, a milliseconds affair, so you can handle a lot more than the number of cores/threads you have available.
cumlord
iirc -5 used 10 job runners by default on that one and it handled fine but i'm not sure if it's comparable
dr|z3d
bump it up to 24 and see how you get on.
dr|z3d
router.maxJobRunners=24
cumlord
i'll try it a bit later and let you know how it goes
dr|z3d
ok
cumlord
with it doing basically nothing (<5kbs both way) one core is maxed out, very weird, but message delay settled down to 2 seconds lol
dr|z3d
that shouldn't be happening.
dr|z3d
by default it should be pretty light on the cpu.
dr|z3d
I'm not seeing much more than 15% of one core with a bunch of stuff running in the bg.
dr|z3d
(reported by htop)
dr|z3d
if you've changed your jvm allocation in the wrapper, make sure your min value isn't too low, otherwise you'll cause garbage collection to work overtime.
dr|z3d
no less than wrapper.java.initmemory=256 I'd recommend, for a router doing a reasonable amount of work (torrents, serving, whatever).
dr|z3d
that value basically tells java "if we're using this amount of ram in the jvm, don't stress, we probably don't need to garbage collect too hard".
dr|z3d
and 256M is a reasonable expectation of usage for a router doing work.
dr|z3d
of course, all bets are off if you don't have available ram to play with.
cumlord
GC could easily be the issue then, it’s a throwaway router I tried to get working under 100M ram
cumlord
Completely forgot I ran into this before when I was trying to see how low I could get it
dr|z3d
if you must less than 100MB, i2pd.
dr|z3d
I wouldn't try anything less than 384 for Java, 512 really is a good minimum value.
cumlord
It was for science of course
dr|z3d
haha, science.
dr|z3d
I've got 16GB allocated here, but that's fairly atypical :)
cumlord
lol most my other ones have 512, the other one on dev im testing had no such problems so that should’ve been a giveaway to me
cumlord
lol what are you doing with 16 🧐
dr|z3d
watching for memory leaks.
dr|z3d
it seems to stabilize at around 1.2G.