#❗-1-18-experimental
1 messages · Page 24 of 1
use simulation distance
jamie pull that picture up
am i gonna need to remake the config files?
the implementation of that may change in the future, but the no-tick config won't be. It'll just be under simulation-distance/view-distance
as the time for a config migration was... a long time ago
thank u
It’s ok we blame leaf on all chunk related problem. 
ye
There currently is no per-world setting for these and the patch to have the server respect the client view distance isn't updated yet either
So if you set view-distance to 10 but the client only has it set to 6 they'll still get sent all 10 and just toss the extra out
so efficient!
there is per world, it's in spigot.yml
iirc we didn't get around to implementing that in CB because we were worried someone would spam the packet to make the server waste a bunch of bandwidth and CPU time sending it chunks over and over
per-world simulation?
Yeah you just need to manually add non/default world like how view distance was iirc
I thought it could only do view-distance
looks at the channel name
Read response to ^
Hey there, today I used the --forceUpgrade argument during startup when upgrading from 1.17.1 to 1.18.1 yet when I joined, the bedrock layer was still at y=0 and there were no new caves generated. Am I doing something wrong?
Yeah, you used forceUpgrade
Roll back to the backup from before you did that and don't 😛
Ah, does forceUpgrade prevent the world from upgrading?
I had tried without forceUpgrade and had the same issue
Pretty sure it prevents the blending with new chunks, I don't think I heard either way on the below 0 stuff but it wouldn't surprise me
y=0 also needs to be bedrock for the below 0 stuff to change
y 0 is bedrock for sure
I'll give it a shot without forceUpgrade
Oddly enough it worked using vanilla but not Paper
Paper recommends against forceUpgrade both due to vanilla issues and Paper issues that have happened in the past
It's not needed and often not tested
Yeah, just gave it another shot without forceUpgrade and still no new generation below y=0
File a bug I guess, you'll probably need to provide at least one region from your world to let others reproduce it
from the 1.17 world ^
Need the level.dat too at least
I'll give it a shot with vanilla again, and then see if just Spigot works before submitting a Paper bug
hats off to thoroughness, my dude 
dose the Increased memory usage when loading old chunks #7094 still a thing or was it patched or dose going from 1.18 to 1.18.1 cause this
we had a fresh world with 1.18
or is the issue also in 1.18
That's only from older versions I think
Not fixed yet, there is a PR but it probably makes doing the upgrade slower
yea
Glad I gave it a shot in vanilla - it seems like my particular world just isn't converting. Maybe I need to start with a clean install of 1.18.
Okay, so one server's world is converting fine, but the other's is not. Not sure how to go about fixing this.
the world was on paper in 1.17? any datapack?
Correct, no datapacks
World was pulled from 1.17.1
It's happening with vanilla though, so it's not a Paper issue
One world converts fine on vanilla, the other world does not convert at all and bedrock is still at y=0
Is there a limit to how much of the world you should pre-gen with Chunky?
Not really a limit - kind of up to you.
If you have a world border most people pregen up to the border.
Ah ok, i figured as much but wasnt sure if there was a downside to generating 25k+ in all directions
ty mate
Larger file size and will take longer; that's a given though
That’s gonna be a massive file
🤷♂️ people love to run out to the border to build, idk why
You can always just do 10k
Does anyone know how long usually new versions go stable
Updates to Paper do not have any sort of estimate for when they release, ever. Any and all updates will arrive when they are ready, and the only thing to do is wait for them patiently along with everyone else.
oh ok ty
I noticed the experimental download warning has been replaced with a slightly less threatening warning. This means its go time? 😉 thanks all of you for your work
Was the average RAM usage reduced?
depends.. Mean I started a world and it was at 3gb compared to spigots 6-7gb.. slowly the ram fills on paper.. highest ive seen was like 26gb for whatever reason.. so maybe restart every 6 hours?
Probably unrelated, but if spigot uses mojang mappings (in which every class and method has a clear name?), why some nms methods are still obfuscated (in paper.jar)?
The final jar does not use mojang mappings because Spigot's final jar does not use mojang mappings.
i cant seem to get anti-xray to work, any help?
anti-xray:
enabled: true
engine-mode: 1
max-block-height: 64
update-radius: 2
lava-obscures: false
use-permission: false
hidden-blocks:
Please send large files/logs to a pastebin
A sensible, modern pastebin. Share text and source code snippets with no hassle.
makes it a lot easier
Is there still an issue with memory leaks causing crashes when upgrading worlds from an older version to 1.18.x as per most recent build #73?
https://github.com/PaperMC/Paper/pull/7097 is still in draft
https://github.com/PaperMC/Paper/issues/7094 is also unclosed
I would hence expect these issues to still jump into your face
May be side effects of the new jvm and it's GCs as well
is render-distance in 1.18 somewhat messing with tps?
My server is just a creative world with redstone and farms. I have view-distance at 20, but lately my ram usage have been sky rocketing. Like 5gb out of 6gb,when normally was at 2gb with farms running.
Any thing that might help other than lowering view-distance?
I cant upgrade plan as it is my own server, and i also wont since i already allocated 32gb to the minecraft network that i have.
i got one yesterday (#71)
Show startup flags
simulation-distance: 10
view-distance: 16
why do things like rockets for elytra sometime take 1 second to deploy on my server. also leaving the nether takes 20seconds.
For those that have started using/tested the 1.18 paper, do you reset your survival world for the new biomes etc? Interested to see how people are approaching, also trying to understand things better.
I had a 1.17 world and deleted all unused chunks to migrate.
1.18 does the rest.
Prediction
Or less technically speaking, low TPS
nah 20tps across the board
its very bizzare
ever since switching to 1.18 performance has been worse
unpredictably
I guess the client thinks that the firework should not have fired for some reason, and the server takes too long to correct it's mistake
its not just that tho
a lot of things are just buggy
how critical would it be if I switch back to 1.17 in a 1.18 world
if the world was generated in 1.18, it would cause a lot of damage
is 1.18.2 paper for bedrock edition out yet?
chunks would get rewritten and your console will die
1.18.2? @shrewd musk has premonitions
Also paper for bedrock? You mean geyser? I don't. I--
Latest is 1.18.1 even including snapshots :)
hey :)
is it normal that if i give my 1.18.1 server (#73) 20 Gigabytes of ram it almost uses 16 all the time? why is my server using 16 Gigabytes of ram even if no player is online?
Have a great day
it's preallocated yes
Don't give it 20gb that's just insane
okay, thank you. I think you're right I don't need 20gb of ram :)
4 should do just fine
Even if no player is online
Mans using more ram than I have in my whole PC to keep the mobs company
Eh 8 to be safe if there is a significant load but sounds like a small server
likely yes
I drop 24 to 64gb on Large Feed the beast servers though but that is another beast 😅
Also that is a big ass server with 128gb of ram anyway
"As you can see, I have rented out this entire warehouse for the ultimate bitcoin mining setup".
"...But can it run an Enigmatica 6 server?"
"...No....."
Nah I bought an old server for 300$ and run it in the basement. Cores are not that fast at 3.6Ghz but I have 24 of them.
Sounds like a Dell R710 😄
pre-ryzen amd cpus most definitely
Yup
2x X5650 if I remember correctly.
i think this is a maxed out setup for R710: CPU(s) 24 x Intel(R) Xeon(R) CPU X5690 @ 3.47GHz (2 Sockets)
Well, at least I can hide behind the fact that I'm not very hardware inclined
To be honest It probably runs better on a modern CPU with far less cores and ram.
And probably cost far less electricity
But as long as Minecraft ist mainly a single-core performance Server, you are probably right
Isn't that CPU shit Tier nowadays?
very
It's a 32nm CPU, lol
the only reason people use them is because they're extremely cheap to find and power efficient
Well a friend just updated to 1.18.1 and the server has like 6tps and when you load chunks in End it crashes and he's unable to get from there. Any way to port him without it crashing?
And how bad is 1.18.1 on a server? 1.18 gave up after few days for him, he had to use backup, and it was fine after that, is it the same expirience?
maybe cause 5690 lmfao
i use amd epyc and i9
not had any issues personally
you can edit the player file directly with NBTEdit or similar to change his location/dimension
Thx
And has anybody heard of the terrible performace and broken End?
Broken End?
Maybe caused by the update from 1.18? Does anybody have any ideas?
Yeah, when End loads, it just crashes after a while
Player gets timeou and it crashes
Has probably been fixed, and if not it will be in the future.
Oh, you're right, new build
But excessive RAM usage has been (partially) patched.
I run my server on core-i5-quad-10GB and I only have issues when zooming with riptide and elytra.
Why downgrade??
.
Sorry
Recovers a backup
So it only laggs excessively when someone is in the end?
No, it laggs when someone exists on the server, it crashes when somebody goes to End
What's the hardware?
Eh, gimme a sec
If you're running on a Celeron or Pentium, good Luck getting that server stable.
But he has 20GB of RAM for the server alocated
cpu and RAM should be proportional.
Only go above 12GB if you have a beefy cpu.
More doesn't mean better.
Ok
It also depends on how you want to use your server.
For Skyblock for an example, you'll never need more than 1GB.
can I dm U timings?
offline mode
that is problem?
ok thanks
yeah not gonna lie you guys blew 1.18.1 "experimental" out of the water
flawless software, ty
i dont see what all the fuss is about 
We just don't want to see people come back crying when stuff crashes and burns
ahhh gotcha gotcha that make more sense
is spawn-limits: monsters: 25 animals: 25 water-animals: 5 water-ambient: 5 water-underground-creature: 5 ambient: 3 in bukkit.yml per-player?
Leave those in bukkit.yml on default and use the spawn limits in paper.yml
is spawn-limits in paperyml per player?
If you toggle per player mobs spawn to true
hmm /paper playermobcaps displays my bukkit.yml settings
assuming it does actually work in bukkityml
then just change both for peace of mind 
yeyee tyty
Is it only marked experimental because not enough testing has happened to the latest versions? Or is it because not everything that is intended to be updated is updated yet?
The second
mobs go brrr
Does 1.18 client work on Paper server 1.18.1?
Yes
What’s the best fork of paper 1.18.1 for larger servers atm
Paper
Recreate without plugin
It's in the same chunk as the source
Could also be the restart that fixes it
what does laggremover do for your server?
create lag
ah they deleted their stuff
i agree, thats why i wanted to learn what they were using it for
Converting 1.17.1 Chunks to 1.18.1 Chunks maxes my 10gb memory right out. I've read this is partially fixed, but my server still keeps dying.
Is there a save way for me to convert the whole map, without killing the server?
it's not fixed at all
no, do it in your machine if your server cant keep up
Fly around and reboot often
It's a vanilla issue in which we've submitted to mojang and asked for some info around the fix
Damn, I hope there comes anything soon from them.
https://timings.aikar.co/?id=5c2f0f121fc84a3eb8afa97ff97fa53a as you can see - Full GCs started. What should I do? Increase Xms and Xmx from 10G to 12G? Or something more?
Aikar's Timings Viewer - View Timings v2 reports from Paper and Sponge
more ram might help but i think its just not possible to have 56 players in survival 1.18.1 without problems
it is possible. I have 20TPS and no problems, only now the Full GCs started becouse space-exhausted
there likely is a memory leak somewhere. more ram and restarting more often can help to delay the issue a bit
do you run a fresh world or is it upgraded?
fresh world. Memory leak? Really? Yesterday I had the same situation. When I had 60 players then I had Full GCs. But after that, when some players quit Full GCs stopped. If it was really memory leak, then the GCs would stop when some players quit?
. leak
leak
if it's a true leak, no, that's the literal entire nature of a leak
if it was just too many players are online and it disappears when people log out, that would be excessive memory usage
so it is no leak
But, I mean, this is an area where people scream leak at anything
if you're having memory issues, go take a look at heap dumps, etc
I think I just have no enought memory in Xms and Xmx.
I will increase it
but should I change other java flags?
aikar has a variant of their flags for +10gb of ram
check in https://startmc.sh/ since the website is basically down (it might have been 16gb +? i dont remember)
Aikar's flags changes when above 12gb ram
ignore the above 12G flags
You're already allocating like 10G of ram
I mean, you can keep tossing more at it, but, I mean, theres very few reasons you'd sanely ever need to go above like 10G of ram
so why now I have Full GCs when too many players? I now that I have view distance set to 8 but I dont want less
heapdump? It will stuck my server for the few minutes, it is too much :/
Then, nobody can help you, glhf
Chunks are twice the size now
either you freeze the server for ~30 seconds max or you have constant lag
take a heapdump if you want to fix it
else gl
Pregeneration is no Option?
wat
How is Pregenerant formed?
new world gen seemed to be working fine on 1.18 paper but running the latest 18.1 it seems to be broken? /locatebiome points me to locations that f3 indicates (and gets generated as) old biomes. haven't been able to find any new biomes on 18.1 accordingly
any suggestions or any one else having issues?
Except rose bush dupe
i suppose a seed change could do it, would that result in /locatebiome and the actual world gen being disjoint?
I was really banking on that
thats terribly annoying
there have been odd reports over the years of random level data being reset
We've yet to reproduce it in a sane environment to even look into it :L
totally fair, honestly more likely something i did rather you guys
Whats the best backup plugin to use?
backups.. while the server is still running and can write to the files mid-copy?
best backup plugin
none
second option is double-writing the whole world state, which sounds worse tbh
This is what save-off is for
> save-all flush
> save-off
$ sync
$ borg create backup::{now} server --exclude server/logs
> save-on
You save-off, save-all flush, do the backup, then save-on
true
Hey I just switched to a dedicated server and whenever I try Aikars flags during startup it just says killed.
Is there any other things that I should install?
are you assigning way more memory than you have?
reduce Xmx
I have 64gb and I set it to 62
(see dmesg, your answer is there)
why...
-Xmx controls the amount of heap memory assigned to the JVM, this does not include other memory used by java, or native memory used by other libraries such as netty (for networking) or SQLite. Please do not allocate all of your memory!
lower it
you dont need 62 gb for minecraft. 8 gb is probably more than enough
Well I have a 64gb dedicated server..
more =/= better
usually start at 6 and go up RN i have my 1.17.1 on 16 and it eats half of it before people login...
I dont want to waste that much
more memory means that there is more memory for the server ot manage
not to mention, that java is a GC'd language, so memory will massively pool up for 0 reason if you allocate so much
So is it even worth having a 64gb server?
no, you will never need 64 gb for minecraft
not for 1 single minecraft server
minecraft is actually pretty bad at managing memory if you give it too much in my experience.
I find 10-12 to be a sweet spot
depends on the amount of worlds and plugins you have
usually some of the plugins and now even 1.18 itself spins off some of the work to other processes
Then why are there flag options for 12+ gb servers
well the update says this
Various background tasks including worldgen are executed on a background thread pool
The default thread pool size equals the amount of available CPU threads, minus one
The upper thread pool limit has been raised from 7 to 255, which should help higher-end machines with improved world-gen performance
threads aren't processes
that doesn't change much
Host multiple servers and sql them
Oops
Pinged wrong person
I mean
i hate that you just used "sql" as a verb
@peak pulsar
this. this very much.
🔫
if you're hosting multiple servers, yea, sure, use a beefed up machine and allocate a few servers on one box

But, that's a specific design layout you've actually gotta cater for and generally a hinderance if you don't plan or design well for it
True
the 12G+ flags are there for the niche cases in which there is actually a use for that
But, stupidly excessive allocations are a waste of space
Ain’t there flags for garbage disposal anyways
(and, with modern MC, I kinda eer against recommending the 12G flags due to changes to the amount of crap int he heap)
Well, yea
better have a good CPU to make sure you can actually make use of that much RAM too; if you use all the RAM and your CPU isn't good enough, then you're just changing which variable constrains you
those flags are there to specifically cater the java GC towards MC
yeah don't ever use sql as a verb again
I will
MC throws around a shitton of memory around the heap and so the flags cater towards tryna keep that short term generated shit out of old gen
but i wana sql my servers
Heh
one hopes one stubs thy toe
lol
🤷♂️
my nick sure is relevant right now
Only once in a blue moon
Okay I see. So the best option to make use of all that ram is to host more than one server. I was thinking of doing that anyways.
Nice explanation of JVM ram consumption.
https://plumbr.io/blog/memory-leaks/why-does-my-java-process-consume-more-memory-than-xmx
Also, I may be wrong, but the trade off with higher GC Times and less RAM is quite bad, as in if you don't have enough RAM so that the GC runs often, you are five bytes away from OOM/100% CPU Load, but that's guessing
And if the GC runs every X seconds is irrelevant, more RAM is most likely wasted, but also keep in mind that RAM grows with player counts (especially if they spread out on the world)
anyone know why chunks are so funky
funky how? are they in a band?
what's your view distance?
vanilla client running 1.18.1? server version (don't say "latest", use numbers)? and again, what's your view distance?
server view distance, not client; there's a bug where the server ignores the client setting
...
just checking/clarifying, somebody was arguing yesterday
what build is your server running?
it should be in the filename
If you're unsure, /version
Most people are running in Pterodactyl where the default setting is to name the server server.jar so I just make them run the command
update to 73 and try to reproduce
ah good to know, thanks
if more than 98% of CPU time is spent on GC and less than 2% of the heap gets recovered, then it will OOM anyways
but yes there is a negative trade off to a lower amount of memory, which is why java uses everything it is given.
but there's also a negative trade off to greater than 10 GB of memory when your server doesn't need that much.
with less frequent but more costly garbage collector spikes, when the server has too much memory.
i see what you mean with unstable, its only been a couple hours and it restarted without saving or leaving a single error line or sign in the previous console log
That sounds like your panel killed it
my what?
do you run the server through a web interface, or are you running it on your own computer via some startup script?
I'm struggling to convert old 1.17.1 chunks to 1.18.1 without crashing the server. Has somebody else the same problem or found a workaround to this?
known vanilla issue......
yeah I know
this is a high priority issue that the dev team is working to fix. like cat said its a vanilla issue
in which we have an issue filed on our issue tracker, and a draft PR to fix it which if you can compile you can use
Thank you, will try it.
for compile this draft PR, I need to manually add this patch into Paper -> patches - > server folder? and just compile?
given that it adds a new patch file, that would work
or you could just checkout the base branch for the PR
that's purpur not paper
eeh? sorry I'm not kinda into this stuff... what is purpur? Some fork of a fork of a fork of a fork?
Sorry if its already been asked, just want to check
[net.dv8tion.jda.internal.requests.RateLimiter] Encountered 429 on route PATCH/channels/{channel_id}
do i need to do anything about this?
nothing to do with paper
whatever is using discord (through jda) is apparently sending so much stuff you hit the rate limit of discords API
ahh ok thanks 🙂
just happened when I upgraded to 18.1 so assumed it was to do with the upgrade
hmm i guess im only one with the chunks
It's a fork of paper, but if you don't know what it is you are apparently not using it.
There was someone with a similar issue here a few dozen messages above
Zyro apparently
I run it on my dedicated server in my basement
it did not kill or restart
only minecraft
Oh nice thanks
I guess he didn't found solution
It started after 1.18.1 update
before no problem
map pregenerated
did it happen more than once?
No. I would have suspected a high mspt, but your mspt looks fine. Is that an upgraded world?
no new one
just once so far
i have upgraded my server ~5 hours ago
had some people online
i did throw in some plugins like discordsrv..
but its extremely strange that there is not a single error log line
Hmm. Maybe take a look at GitHub and see if there is an issue about chunk loading already. If not make sure you are in the latest build and create an issue, that definetly is not how it should behave
does it have the initial world load even?
people were playing for ~3 hours
ok thx
Vanilla lol enjoy bad performance
Do Viabackwards And ViaVersion work fine atm?
prob best to ask the via version discord
but I know they've fixed a couple issues since their initial 1.18 release
i enjoy servers not rebooting and deleting progress
it just happened again so now yes 😛
and there's no logs whatsoever? how is it restarting? that's not a normal server behavior, it should just stop if it breaks or you tell it to stop
it's not OOMing is it?
uh
(Out Of Memory)
im using Xmx7G and i have 8
Lower it to 6
on debian
It's not writing logs at all? Are you sure you have the right file permissions?
its writing logs just not when it crashes
see dmesg
i will try 6
Doesn’t sound like the issue honestly, but might as well try it.
😎
yeah try 6
til headless debian fresh install needs more than 1G
it's not just the headless install, it's also the JVM overhead
it's not debian
the minecraft server / jvm has an overhead
lowering it to 6 will "fix" it
anyone run into bugs where player items instantly despawn on death?
Recreate without plugin yet?
It's only just happened twice. Only plugin running is dynmap.
Oh yeah, wondering how ViaRewind/ViaBackwards would work now. Would they just give up? Would they not send blocks outside of Y=0-255 and add a fake bedrock floor? Would they use two dimensions for the overworld, that you transition between automagically?
No blocks outside of 0-255
Ah boring solution :(
Two dimensions would have been dope as hell
Considerable overhead tho
not really solvable without basically oddball teleportation which would just ruin the feel, etc
Bedrock roof could also cause client mod breakage
bearing in mind that worlds can be even higher than vanilla
But doesnt the new 1.18 generation cause a lot more of the valuable ground resources to be below Y=0? 1.17 Via-players would just not be able to access it
Also insane prediction issues caused by players walking into the void, I would imagine
well, you'd need to dynamically teleport people
and resend the entire set of loaded chunks everytime you moved between groupins
translate and convert every position related packet, etc
I guess it doesnt make sense to do all of that for what the plugin aims to accomplish
It's not really meant for SMP primarily
What is the point of the viabackwards forwards plugins anyway? Am I just missing the point, surely it's just as easy to update the client?
YOO HELLO
its for people whos machines are not powerful enough to update
or people with mods that havent updated
Heloo, aha I see, does seem with 1.18 they will get 3/4 of a game though
wut
Well, under 0 and over 255 would be forever unavailable to them
Lobbies
You can have a second lobby for old mod packs, or... viabackwards/rewind
I cant imagine people wanting to PLAY with an outdated client
well people do
there's still 1.7 and 1.8ers out there
Not on 1.18 servers
Viabackwards/rewind allow outdated clients to join newer servers
Those people would play on native 1.7/1.8 servers
Dude do you use any plugins or java arguments?
If yes get rid of java arguments and plugins that somehow manipulates with chunk loading
damn
so this is interesting
i set a world border of 30kx30k
rendered all the chunks
turns out it rendered 80kx80k
now my world file is 300gb
players alr started playing on it
heh
is there any fixing that
or do i need to pay ovh for larger storage
Whatever you used to fill should have a trim fuction
Does no tick view distance still work with 1.18
No it's sim distance in 1.18
No, only vanilla's simulation-distance (for now)
Ahhh
Same idea just different names
Ahh so it doesn’t matter what my no tick is should I disable it or does it matter
Should I have my server properties view distance at like 8 and the sim distance at like 32
I would not set sim distance higher then 10
Maybe put view distance higher if you want like 12 or if go crazy 16
Keep in mind setting view higher means you're sending more data to players Wich might have negative impact on those who have a slow connection
I've noticed 1.18 is just slower in general. Mobs seemingly have a bit of lag when moving. They'll appear to "jitter?". tps is 20. I don't have a lot of upload bandwidth, but even when I'm idling alone, it seems like that
If anyone has any ideas (or knows why), let me know. 🙂 I can provide more information if it is needed
i'd say "do a backup" but lol
it probably wont though, a trim is just deleting files, worlds are not saved as one large file
worst case re-fill if it under cuts by a bit
I tried using chunky but kept getting random chunk corruption. :/
the worlds are currently on a raidz2 array of HDDs. I don't think an SSD would make that much of a difference for 2-3 people playing? I haven't tried it.
provide timings
What's that?
/timings report
thank you. That's new to me. https://timings.aikar.co/?id=22f40dc7ccf645b9b158a4f93db92a1e
Aikar's Timings Viewer - View Timings v2 reports from Paper and Sponge
I see red values. Probably bad x)
I'm sure I have something configured wrong somewhere. Although, I don't think my 1.17 had this weird issue
https://paste.gg/p/anonymous/355814a213d948e1b38c1318d37a3f77 paper.yml
probably one of these plugins that may have bugs in their 1.18 version? i don't know
also, newly generated worlds. Did not carry over from 1.17
is paper 1.18 out stablely
like that bottom one
is it a stable release
or what
at this point it's pretty much just incomplete
idk if they labeled it incorrectly on apex
there are two main bugs remaining that you'll see, 1. a memory leak when upgrading old chunks, and 2. lectern hitboxes are broken
so if you're upgrading a 1.17 world, I'd hold off for the moment, there is already a fix proposed for that first one although it has flaws, so it's possible another fix will be used
lectern issue isn't related to upgrades
stable wen
when you stop asking and read one screen above your message
seems like mobSpawns are taking up a large portion of your tick (9-10%)
change your ticks-per for monster-spawns from 1 to 3 or 4. this will lower your spawn rates. also your spawn-limits for monsters is set quite high. I'd recommend lowering it from 70 to 50
less mobs spawning at a lower rate will help you out
very much appreciated. I'll take a look at it and try your suggestion
you have tps loss because you have over 500 entities for only 2 players
see?
I see. thanks. I don't think I changed any configuration between 1.17 and 1.18, but alas 1.18 is quite a different beast I believe
but yea, other than that i didnt see much
thanks for looking into it
Good idea
I'm at 1.17.1 is it recommended to upgrade to 1.18.1 currently?
its up to you honestly
looks at channel name
through Apex hosting
Unless you're starting a new world it's definitely not ready
it seems to work fine for many but it runs slower and you need to keep regular backups
Ah I gotcha. I'll wait for the stable release. Thanks!
There is a memory leak (it's in vanilla too) on upgrading old chunks
You can kill a server with one player zooming around
The speed issue is likely just a 1.18 thing and not a 1.18 paper thing (considering it runs slower when playing a local world too), so stable likely won't fix that issue in comparison to 1.17 haha
There are some major perf patches from 1.17 that aren't updated yet
They probably won't be before it's considered stable at this point
that makes sense, but I'd also suspect that 1.18 will also just never run as fast as 1.17 due to changes without significantly more optimization on top of what 1.17 had. 1.18 is just a heavier release
Although those patches are more likely than usual to cause some nasty bugs if there are any issues with them
So if it's marked as stable before they get updated you still have to look out for them later
Not that you should blindly run any build of paper
just noticed, my xray config wasn't updated
should I move the old paper.yml somewhere and let paper generate a fresh paper.yml?
@jade gorge well, you can see the one provided here:
Anti Xray configuration guide: https://gist.github.com/stonar96/ba18568bd91e5afd590e8038d14e245e
oh, thx
You can if you want to get rid of all unwanted stuff
Just manually paste stuff back
Are the backups just precautionary or is there still a (higher than normal) chance for the world/files to get corrupted?
Yeah which is why I added “higher than normal”
given our position in the Mojang dev cycle, I think the answer is "yes"
expect 1.18.2
has the anti xray guide been updated for 1.18?
yes
Any updates for the 1.17 memory leak?
should i make the max-block-height higher for the overworld anti xray in 1.18? or not really
oh i just saw that it says to do that
I mean, the issue tracker is where you will find them most likely.
I don’t remember the issue link
I know there is one with entities, although it hasn’t been triaged yet.
Well there is one currently that is titled something like “memory leak with hoglins”
And then there’s a 1.18 one with chunk generation
Ok it’s not closed yet
quick question, what does triaged mean?
Anti Xray configuration guide: https://gist.github.com/stonar96/ba18568bd91e5afd590e8038d14e245e
In what context @stuck mural
Well i was looking on the issue tracker and i will see the label Needs Triage and I dont know what that means
Basically means the issue has not been evaluated yet, e.g. is it reproducible etc etc
ah gotcha. that explains it to me thanks!
not me asking when things are ready. plz no pitchforks.
did the papermc.io/downloads page change from experimental builds to early builds? or has it said early builds the whole time?
yes that was changed a bit ago, the two main known bugs yet are a memory leak in upgrading old worlds, and a broken hitbox for lecterns
but other than that it's mostly just unfinished
I use 1.8.8 papemc so how I protect ?
1.8.8 has been updated
So
just get a new version from https://papermc.io/legacy
correct
Yeeee
nice! ty
just note that 1.8.8 is still not supported, and no future updates will be provided whatsoever
it's what's holding us back as a world really, world peace can never be truly established when everyone is so divided over pvp
throw the 1.8 people in a volcano, now you have world peace
idk if it's unique to paper, but whenever i walk into a lectern the game glitches out, all my players have mentioned it.
bugged hitbox with lecterns currently
AFAIk anyway
okay ty for confirming
1.8 pvp is so nice though
into the volcano
Brain numbing
I find it more competitive than 1.9 pvp but that’s probably because I used to main games like SkyWars when I used to play mc
clicking
the spam clicking
1.8 papermc dowload ?
No support now or ever
the new pvp was the dumbest thing mojang ever did
even after all this time i refuse to pvp on it
How on earth are we talking about fucking 1.8 in the 1.18 channel?
Can we like, not?
it has an 8 in it!!
Where can I find Minecraft alpha 8 builds?
ah yes the java bug
The log4j exploit has been fixed. The builds are not quite considered "stable" yet.
Updates to Paper do not have any sort of estimate for when they release, ever. Any and all updates will arrive when they are ready, and the only thing to do is wait for them patiently along with everyone else.
Ty
You can use it, but it is not recommended for Production uses
I mean, there is a plugin for old combat mechanics.
Wasn't 1.9 the version that added the features that let OCM work?
Like, the same time they changed things they made those things configurable so you could put them back to the old way
big one was the attack speed iirc, but, that was already there, just, it worked more as like a warm up type thing iirc
if you kept attacking too fast you wouldn't deal as much damage, but, it was less visual and noticable when you're just going full click fest anyways
Sure 1.9 added the attack speed stuff but it also made that something you could configure
afaik you can limit the sweeping stuff on swords too
As they changed things in 1.9 they made them configurable with the attribute system
I mean, I had the attack speed floatin around like 6.3 and most people seemed to be happy with it
there were still some changes but overall, it kinda worked well
The only thing I remember people were saying you couldn't reproduce on later versions was the knockback thing where it's dependent on when you joined the server
But spigot and paper patched that anyway afaik
So there has never been a reason to stick with 1.8 unless your server kept a pretty blatant cheat enabled that made the already joke PVP completely unfair
I dont suppose anyone nows how to disable the shulker dupe patch. i looked throw the paper.yml i couldnt see anything to do with shulkers or containers
Why does paper 1.18 have like 50 tons of fog? Is there any way I can reduce it
Update your client to 1.18.1
Ohh is fog a client thing?
It's based on the server view distance but yeah
Ohhh ok thank you 😁
@tawdry spoke advice your player to use fabric and sodium extra to remove it completely 
In 1.18.1 they pushed the fog back a bit and made it cylindrical so you can actually see the ground again when you're high up 😄
Ok thank you 😁😁
Also make sure you don't accidentally have a really short view distance due to old no-tick configuration
Does that make the simulation distance the same as the render distance minus 2 or something
hmm, I probably would have done that on a creative server I used to run
Honestly not a bad idea
We gave up on minecarts and just added warps and mostly didn't want people to build redstone things
Although I suppose it would make placing liquids impossible
true. they will just be staying whereever you place it forever
Can we set it to like, 0.5?
I dont think so 
0.1, only the block you're on top of is ticking
is it safe to start a new world on paper 1.18.1
Yes
No, your server host will implode
Fake News
I heard it from Jeb, trust
is it known that lecterns' hitboxes are broken?
Expected behavior Should be able to walk over or on a lectern. Observed/Actual behavior See video: https://m.sjh.at/2021-12-06_18-07-00.mp4 Cannot walk across as the player is constantly teleported...
^such a weird bug, always freaks me out whenever I walk over a lectern in my trading hall
is flying into an unloaded chunk and dying of kinetic energy normal? with prevent flying into unloaded chunks on
The joys of map generation.
Guys, can log4j issue make server unsafe if 1.17 or 1.16 users join server using, via version and via backwards? Although i fixed the issue on server
if the server is up to date, it's fine
iirc all clients from 1.12 and up are fixed as well
At this point, all clients are safe though. Mojang released updates to all of them. So unless you’re using like an outdated third party client, clients are fine
how many cps?
one
Well, 12 cores and 24 threads, only using all of em for original generation, gonna let it run for the night, should be done by morning
..how much are you pregenerating?
-5000 to 5000 each direction
and then also generating a map for it, which is the actually intenstive part
dynmap?
yeah, nothing else has as easy of an interface, but it creates an absolutely staggering number of files
if you do know better let me know
it does. had to shut it down; to huge
nope, best plugin for that job i stumbled upon
Bluemap is amazing and i find myselfs map having a smaller size blueprint too
Yeah, i also use bluemap but it's not as convenient because the zoomed out view is so... abstract
True
I have space and processer to spare though, so i don't mind
once its done generating everything the first time, everything else uses like 1/10th the resources
my biggest performance drain is when people are generating chunks on the fly and this just completely negates that
Minecraft moment
interesting, the entire east side of my map is generating as some super cold biome variant
what Plugin u are using for pregeneration?
chunky
so far things seem to be going well. I kept a lot of my old pre 1.16 generation where town and such is, yeeted a ton of old areas that no one had utulized with MCA editor, and now i'm generating new terrrain.
current progress
and yeah, that thing somewhat above me that's weirdly square? 1:1 original zelda map that someone built on my server about 2 years ago that i'm never gonna get rid of, cool stuff
anyway, off to bed because this is gonna take ages and i really need to sleep
I'm planning on using mca editor too
good to see everythings going well
what are the chunk borders like
So, the 1.18.1 version is available through the hosting service Apex now, does that mean it is stable enough to update the server I am hosting?
Idk how apex decides when a build is good, I doubt they do testing ^^
Our stance is that these builds are still early, but you can use them if you test your setup yourself and confirmed your daily backups work
(there are over 50k paper servers on 1.18)
Is 1.18.1 working good for y’all
Yes it is
There's an absolute ton of chunk borders in that picture, i'm talking like.. literally hundreds in all sorts of weird ass shapes
some are kinda obvious on the map to the left you can see an obvious straight edge that used to be a world border
however, in game, you can't tell its a border at all
or that it used to be
Hello, how useful is using Swap with paper on NVME drives.
I'm making a small server for me and my friends and I can choose between either 1,5G pure ram or add whatever amount of Swap I want (we will have the chunks pre generated with chunky).
Swap will obviously be slower but I'd like to know if anyone can tell me how strong the impact will be of slower ram vs less ram.
just swap will run horribly
1.5gb normal ram and like 2gb swap will still be pretty bad
Not really worth destroying the NVMe drives.
yeah, that would probably be terrible
on the other hand, i've noticed that using ramcache through primocache doesn't really do much to make my server any faster. However its already hosted on a PCI-E Gen 4 SN850
ok, actually leaving now, goodnight all
Hmm I see. on 1.17 it ran incredible using paper even on short amount of ram but now the chunk generation is a lot heavier and caching is really useful there so I guess i'll just have to try it out
once chunks are done generating things will become smooth again
we will have the chunks pre generated with chunky
so this should help right?
yep. That's what i'm doing right now myself. (Scroll up)
but i think chunky is kinda brokwn now
it's working perfectly for me.
idk if mem leak is fixes yet
I see, well that's what I'm gonna do to. Hope for the best. Gn
Ill watch out for that thanks
It has gone up to 16 gigabytes of mem usage for me, but that's something I was expected
goes back down to 2-3gb once its done (and my maps are still also simultaneously being generated)
it mostly only happen on 1.17 worlds not new ones
I see. thats good we plan to start a new one
oh yeah, if you're gonna use MCA selector to destroy parts of maps, you can't open maps that generated a combination of both pre 1.18 and post 1.18 terrain
it freezes at loading the map
what kind of specs do you have?
2.6 xeon and 6.5 g mem allocated
I see, well I won't have that kind of processing power ^^'. That'd be too expensive for this small project.
Tho i guess Ill just have to try a few tings to see what works best for my hardware
i use a vps for 10 pound a month
pregen seems to be popular though (scrolling through the channel) with the more complex chunk gen
10 pound a month for a dedicated xeon huh? where do you host?
time4vps
idk if dedicated
haven't heard of them yet
I wanted to host on Hetzner though they do give you a lot of ram huh?
ye
Ill check them out for sure, looks interesting. Thanks!
i chose 8g left 1.5 for linux
Please do not use that
It’s not make for Minecraft
To get optimal performance you want a host that specialize for Minecraft
if you use time4vps you have no rights of complaining about lag
And from hetzner I'd recommend getting a dedicated one, recommended from me is the AX Line
from 40€ 👀
True. Have it too
Yes, but then you can even host multiple servers and cant complain about lag
As it is fluent af
If you just have a small server
Obviously a more expensive server will have better performance but that is just too expensive for 4 people
Oh then...
But thanks for the input 🙂
Maybe look at the ones eternity named, or look at sparkedhost
I would prefer a linux machine even if it means more work for me since I can also host other related thinks on there.
i didnt
i get no lag
Oh well... I guess I'm just a bit frustrated what I didn't get the performance I used to with 1.17 but that is how it is now. World gen has gotten complex--
also i use puffer panel not screen
Anyways thanks for the help, lots of different opinions here I'll need to check out some stuff-
Is it safe now to update on 1.18 ?
there is a memory leak when upgrading maps from pre 1.18
otherwise, it seems stable
(bar a bug or two on the tracker)

I'll probably go your route but with hetzner. they offer simmilar pricing and I already have other servers there. I'll just rescale until its fluid and hope its gonna be earlier then later. thanks 🙂
nice btw use pterodactyl or pufferpanel
better than using screen
what's that?
try heavynode if you opt for non-dedicated shared
they're one that doesnt oversell
it gives you a panel with console and file manager no need to ssh or ftp
every time
everyone just recommends different hosting providers xD
Ah I see. sounds interesting though I don't see anything wrong with just setting up a systemd service and sshing sometimes. Might be convenient though-
yea nothing is wrong but it gets too anoying to ssh everytime to start or alter config
you can just go into ur website for this
Their pricing model sure is interesting
true
what about it?
there aren't many good shared hosts, heavynode just happens to be one of the most transparent ones I know
Well you pay select a hardware class then just scale ram.. not used to hosters specifically for minecraft- maybe thats normal
that's normal for shared hosts yes
also take your time and keep in mind some hosts like ovh ask personal info like passport and certificates
the key factor that makes or breaks a shared host is whether they oversell hardware
can i use the early build for making the server, and wait until the fully stable version releases then have it go in production?
so i'll use the early builds for 1.18.1 as of right now downloading plugins setting everything up etc then update to the stable version whenever than releases
so just to setup the server
yes 👍
protocol translation is also an option if you don't care about new features but do care about stability
Is paper going to bump log4j to 2.16.0? It's not a requirement to fix the CVE, but Apache still recommends to update
when it's released, maybe
Better safe than sorry
excuse me for being a little bit ignorant... but what is the difference from an Early build to an Experimental build? it means it is stable?
Thanks
maybe early has not all the functions and Experimental has it all but untested?
just guessing
they seem stable, known bugs but aren't "critical", etc
well, except for that fun vanilla world upgrade one...
2.16 solely hardens 2.15's CVE
Oh, thank you! i understood that is more stable. i'll wait a little bit more until a stable build is release :3
Thanks again! n_n
Did some upgrades at work earlier but eh, nothing major tbf
Work is still a mess here
Majority is log4j over slf4j or some weird in house projects from years ago wrapping around souts but eh, luckily logging hasn't really been a (well used) thing over here
I cant find the 1.18 tracker.
@modern lintel
I'm getting tons of these errors, it's safe to ignore it?
https://paste.gg/p/anonymous/10f6e66007174dbbbce257b75e6c1e3a
Is there any paper build for 1.18.1 that has chunk-loading section?
no
runs
That has what?
I assume the chunk-loading settings in paper.yml?
Hi, I am having problem with my 1.18 server. I have an 8gb server with a few plugins. (Only essentials is not officially supported) But with around 6 people it go's up pretty fast to the 8gb and crashes. Spark saying I only use 3gb I find it weird there is so much allocated memory. Is this a known issue?
Do you use a panel?
Yes, pterodactyl
Can you send your startup flags
java -Xms128M -Xmx8192M -Dterminal.jline=false -Dterminal.ansi=true -jar paper-1.18-66.jar
This is set by my hosting company
Normaly cant change it, but I work there so I can
What host do you use?
Uhh we are getting an error sometimes when players join (its not consistent at all)
Internal Exception: io.netty.handler.codec.DecoderException: java.lang.IndexOutOfBoundsException: Index 54 out of bounds for lenght 5
There is no errors in the server log, and there is no errors in the player's log what could be causing this?
Ask them if they can set both xms and xmx to 7b ram (not 8)
okay, what will change?
Server shouldnt crash like that so they might be overselling
okay, Ill try thanks alot
np, if they say no and you dont want to change host ask for 2gb swap ram
actually that might not help if they oversell
Well, I work there so I can change it myself. I personaly dont think they oversell sinds I can see that I have 8gb allocated to my server, but I could be wrong
we only know that it happens to every player trying to log in, after waiting for a while it stops
version bump to l4j 2.16.0 would be nice for hardening 👍 i know it's not critical but it's a drop-in replacement anyway, so
cheers for your work guys, the paper contributors are awesome
Is there a known bug in Paper with Portals?
There is some nasty issue with memory in 1.18
Who?
Aeyesi
Yep I did with --forceupgrade
ah
oh okay
advice has been to not use forceUpgrade
what? xD
the startup flags are from 128m to 8gb which is bad for performance
I got 12GB XMS/XMS and its buggy anyway
i think 1.18 just needs more memory because of larger chunks
12G is more than enough I assume
depends how many players, what they are doing and what plugins
There is a memory leak in vanilla
ah didnt know, just restart your server pretty often i guess ¯_(ツ)_/¯
Can anyone Tell me if there is a known bug of nether portals? Players on my Server Are spawning somewhere in random Locations, instead at their own portals, Any known solution?
Well, just opened Saturday and some players are already a bit irritated from all the crashes and restarts
lol
You may want to reduce your Xmx a little bit, I set my Xmx to 10G but it actually uses 10.9G.
-Xmx controls the amount of heap memory assigned to the JVM, this does not include other memory used by java, or native memory used by other libraries such as netty (for networking) or SQLite. Please do not allocate all of your memory!
tl;dr -Xmx is mem for the JVM only, not java in general
do not use --forceupgrade heh.... well that was the first thing I did
If you like the cool looking edge on your border it’s fine /s
I'm going from a 1.16.5 world to 1.18.1
is there anything special I need to do other than change the jar file
Nope
Other than ensure your backups work
On mobile too!
very quick yes thx
It’s your view distance in server.properties
Can I make that higher without setting the viewdistance to it?
Since paper/spigot has its own?
ah
Change accordingly
whats recomended sumilation-distance?
Fog is kinda fixed in 1.18.1. It doesn’t make your server looks like silent hill anymore but is still kinda meh. You can remove it on client side with fabric + sodium extra
10 is default but the ideal value is made based on your hardware so no way to recommend. Test and see
genuinely surprised it's this even
Y’all doing vote? I just update it 
yeah @neon shard just update it 😛
Wait what, that end bit. Is paper stable for 1.18.1?? I thought it was still unstable
I just Updatet over night. The next morning it was like 1.18 in your faces
Updates to Paper do not have any sort of estimate for when they release, ever. Any and all updates will arrive when they are ready, and the only thing to do is wait for them patiently along with everyone else.
same lol
Honestly the bugs are, from what i've done so far, pretty minimal, outside of a couple weird underground light rendering things
however now that my map is completely pregenned i think that'll be fine.
NO FING WAY i was humming giorno theme while clicking link and boom
I'm willing to bet paper will be marked as fairly stable sometime very soon anyways :P
It's already marked "early" and not "Experimental" so take from that whatever you want
👀
Don’t forget it’s a less aggressive color now
I mean, looking historically from previous updates, it's almost the time paper gets marked as stable anyways. Plus, 1.18 wasn't the biggest update ever; the biggest change is just generation.

1.17 was marked as stable from build #80 i think
all joke aside just make backups

