193 Replies
@Livor Mortis
wait what
I didnt see the
> Killed before
idrk what that's about tbhyeah it kills the server automatically
ended up duping items apparently
spark
spark is a performance profiler for Minecraft clients, servers, and proxies.
Full report
see if those make a difference
All tick issues
Everything else is good
yeah i was about to say
i dont have
Any documentation for switching from paper?
just change the .jar file out
same everything else
ah, what about paper configs?
wdym
🤦♂️ I know what they are
I mean what about them
do they get replaced ig?
no
pufferfish uses them
pufferfish is just a set of patches on top of paper
so all your paper settings etc are saved
its called not using contabo, they oversell a shit ton, networking is shit etc
I paid for 3 months 😭
that too
yeah ill switch
eventually
cut your losses imo
oh also disable tps catchup in pufferfish's config
makes player experience better (imo, anyway) and the server has to do less cause its not trying to catch up after a tps drop
ok
pufferfish > purpur?
yeah
mfm doesnt support 1.21
just look at older builds
or update
1.21 isnt a good version.
it has dupes etc
i mean i mrunning p much
a base vanilla servert
would recommend updating to the latest pufferfish build
did
should just work, but do take backups
oh
as in build 27?
seems like it
i might eventually
servers boutta make it past the two week mark lmao
did a bunch of optimizations: was great for a while, then everyone started coming together and TPS shit the bed
cpu
i think is the issue
this looks fine tbh
if you have more than a one-off lag spike you can get a spark report and post it here
@Livor Mortis I literally cant find a reason
I'm trying to figre out why this shit's going on. I keep constantly having TPS slowly drop more and more and then it crashes and kills the server
i limited entities down to 70%
i would offer money but id ont know if theres any way to guarantee af ix
i was thinking of upgrading my contabo but theres an off chance thats the issue but i dont know
spark
spark is a performance profiler for Minecraft clients, servers, and proxies.
errr
right before crash
yeah contabo
I know
:YEP: well uh
sorry for your loss man
should i upgrade
to a better contabo VPS
w/ more cores?
how much do you pay
5 a month for 6gb and 4 cores
100gb nvme ssd
what region
na-east new york
this is really helpful
you can go with pebblehost 1gb/$
whenveer loaded entities and loaded chunkks go up
the tps goes down and it never decreases
they have a i9 9900k for the 1gb/$
does it make that much of a difference
big difference
i remember when my place went from bisect's shitty xeons to a 7950x
it literally boosted us by 20+ players (400 mod modpack, that was optimized to OBLIVIAN)
from like 15 players
its a big difference having a more powerful cpu
I thought minecraft was a lot more multirheated?
if you wanna test i have a EU i9 9900k to see its performance
not even a BIT
minecraft has never been multi threaded
never will
mojang and its team wont ever make it as itll probably take a entire rewrite
i thought it switched after 1.12
nope
ah
so you think that by switching to pebblehost, everything will pretty much fix itself performance wise?
pebblehost, you can disable the ddos protection
itll give you a boost
it might not Fix
but itll help 100%
again, i can give you a temp instance on my own machine that i run in the EU if you wanna test
as it has a i9 9900k like this bisect plan
would Folia help at all
no
as folia needs a hella lot of cores
ah so its intended for like
100+ ccu
like 16 cores minimum
or reccomended either one
its best for that cpu you have rn, but if you had access to it 100%
ah
all cores, all ram, etc etc
but i suggest pebble
they are pretty good and reputable
i had used pebble in the past with no issue
i had 3 days downtime once
really annoying
it was my node i think was just gone for some erason
support didnt help at all
:GhostHuh:
how long ago was that?
left a sour taste
last year actually
october
I know what's wrong with it. It ain't got no gas in it
LOL
lower your xmx my dude
fuck i didnt even see that
Lower to 4?
something like hat, I'm not sure tbh
is that the issue you believe?
yes, if you're server is out of memory, it's unhappy
bc im normally hitting a max of 10 players
maybe add some swap
aikars flag, 4.5 good?
i still think swapping would be better :3
no
you're only using like 1.3GB on the heap
or something like that
so go down to 3.5?
yea
if you wanted to go down to 2gb or 3gb, you would probably be fine
this is the issue with vps's
no overhead
do you think 3.5 is overdoing it
on my plan analytics i think
add some swap
it had added some
1 sec
spark says 0 swap
yeah my ram maxed out at 2600
with 11 players
that's before gc ran
if you look at what memory usage was after gc, I think it's only at 1.3gb
ATM its using 2.3
3 players on
aikar's flags will only run gc at like 40% memory usage or something like that
@Snow Kit this worth enough to ping dan over?
so keep that in mind
¯\_(ツ)_/¯
no clue
i dont know if i have any logs
fair enough
of it
it might have been discord saupport instead of hteir tickets on website
if you can find it, make another ticket
i dont remember
if they are the same we will ping dan :thumbsupcat:
server killed itself lmao
ty. i reduced it down to 3.5 to give it some overhead (2 should be enough for the ubuntu OS and other stuff)
it also depends on just what plugins you're using if they'll need memory outside of the heap space
it's kinda hard to predict how much overhead you need tbh
voicechat might be a big one
not really
geyser is the big one imo
along w/ floodgate?
and broadcast too
soso roughly
ig
ajleaderboards will really do some funky stuff with memory usage outside of the heap
dont have ajleaderboards
its p much a basic vanilla server
ik, I'm just saying that, I've seen it a few times
i think the memory might be the actual issue; i would have never noticed that. didnt know how to navigate spark reports
if so theres no reason to upgrade my memory
yeah, if your mspt is well below 50, but your tps is low, then it's an underlying issue with the machine
is the i9 9900k on pebblehost native or
their minecraft shared hosting is docker/pterodactyl based now
i ssh into a protected ubuntu installation
i wouldnt call it "Protected" but sure
pebble is probably a bit more secure
its prolly equal
pebble would probs be more secure, contabo has.. its own issues
1 more thing: since the amount of entities seems to directly correlate with degrade performance, should clearlagg help it out?
ive tried touching up spawn rates but it still builds up over time
red-entities
green- tps
purple - mspt
yeah
thats also due to the bad cpu
sorta
ok wait. do chunks and entity unload by defualt?
because entity count never seems to dissipate
I am aware of the tree chopper stuff but
I'm wroking within my limitations to make it playable
Reddit
From the admincraft community on Reddit
Explore this post and more from the admincraft community
Very similar to my issue
yeah
they do
on my graphs, they never go down
plan & sparks
if you are constantly in 1 place, or moving itll be roughly the same line
i meant even if like you have 7 players on
and it goes down to 3
its still the same as 7
after how long
almost as if those entities are having a jolly fun time together
no matter how long
odd
wait you have the Plan plugin installed?
thats pretty cpu heavy iirc
It defo helps with this stuff however
im only using 9% of my cpu
try disabling it and seeing its performance diff
i was having the same issues before the plugin was insatlled
nvm then lol
ive had it since like day one really
10 players together
bunch of entities
and loaded chunks
render distance is low
same for sim distance
i just told players to use bobby and DH
so its not server impacting
true
good ask of em
even when no ones exploring
like theyre going to one central place thats already loaded
its just slowly loading in more and more
idc abt posting this no care in the world about security
ill secure it another day
im just posting so you can have access to the graphs because i really really dont know what to look for
are entities more of a cpu thing like calculating their ai all at once
or memory like storing abunch
if its proven to be the CPU then ill switch but
i dont want to jump ship if its not
something like that yeah
so adding on another 24gb wont do jack
:PepeCough: i can offer you a test instance
third times the charm :EZY:
maybe
but not greatly
like
wdym by that
how bad your cpu is
is
Aternos has better cpu
gonna be honest i blame my buddy for putting me on contabo
kick his ass
he uses it to host his search engine
ok sick its down
i got a EU i9 9900k, and NA 9900x servers for my own reasons. not a server host
so if you wanna test the cpu's i can give you a temp server with the same ram and storage you have rn for contabo and test it out
i have 0 idea why i wanna let you test so bad
i would maybe like to try that out but i dont know well it would translate consider id need to reconfigure just to make sure utilization is the same
but i guess i just wanna use the mfing server i pay for, for once lmao
either way we dont have 10 players to join
to increase entities and chunks
drag -> drop -> small fixs if server breaks
problem solved
where did you buy your servers
hetzner, and privately owned
hetzner is for EU so the i9 9900k
Dont use contabo
its not for games
they also overshell the shit out of their servers
wow i love this
Wow 10/10
ive brought a pebblehost server
you shouldnt really rely on systems like this
instead, when you get a lag spike, just peek at your spark report and see what is actually causing the lag
entities
its 100% entities every single time
and i think i misconfigured it
because it doesnt go down at all
this is not a good idea 😭
then use an entity limiter, dont periodically purge them
Pufferfish also has a setting called DAB
You could consider using that, it helps with entities quite a bit
entity purging is a bandaid fix at best
contabo vs pebblehost
pebblehost at end after last tps drop
What am I looking at here
constant drops were contabo
pebblehost
looks about right