Dam, ok I got it.
I was doing ```java
player.sendData(Main.PLUGIN_COMMANDS, out.toByteArray());
instead off ```java
player.getServer().getInfo().sendData(Main.PLUGIN_COMMANDS, out.toByteArray());
face palm moment
#waterfall-dev
stark bone
stark bone
Also had to use a BukkitRunnable for teleporting so it delays and tries a few times to make sure player is in the world
ornate jasper
Did you fix it?
If not, i believe you have to send the data to the server, not the player
Because you're doing player.sendData
coral lichen
any suggestions to injecting code here via a plugin?
Having to block packets from reaching other handlers
sly crown
that's an xy problem if i've ever seen one
trail plume
I'm planning to add an event around there
sly crown
"how do i mixin bungee"
trail plume
Potentially
coral lichen
any solutions for current versions?
trail plume
Not really
coral lichen
maybe get the Channel after server started and inject into pipeline afterwards
general requirement is that i need to do it before player connection starts, so that's my issue.
trail plume
handshake is basically the only reliable place you're going to be able to do that properly
plugins load waaaay too late for bytecode transformation to be realistic there
coral lichen
having to do a system to filter packets
since people get layer 7 attacked with bad requests and bots and whatnot
By doing it as soon as possible, i can withstand a larger throughput.
also PlayerHandshakeEvent can't be cancelled
trail plume
You can close the connection there
coral lichen
https://sfy.cx/ss/u/7775.png
If i close the connection, wouldn't it keep running the code in the switch?
coral lichen
yeah, that's the thing
trail plume
God, I fucking hate lombok
coral lichen
hopefully Thread.currentThread().interrupt(); wouldn't screw it up
ruby dust
@coral lichen you can save old SERVER_CHILD ChannelInitializer, replace with yours one, do some stuff then call initChannel from saved ChannelInitializer
coral lichen
yeah, good idea
bleak current
does anyone here happen to know how to send an empty packet?
so that it triggers bungee's empty packet exception
ruby dust
writeVarInt(0, buf) ?
spring bloom
write an empty buffer
bleak current
yet when i send that then bungee says length is over 0
probably doing something wrong in client
real cave
Make sure you have compression enabled
Then send a compressed packet with zero payload
bleak current
Channel ch = ((IClientConnectionMixin) client.getNetworkHandler().getConnection()).getChannel();
ChannelHandlerContext ctx;
if (ch.pipeline().get("compress") != null) {
ctx = ch.pipeline().context("compress");
} else {
ctx = ch.pipeline().firstContext();
}
PacketByteBuf buf = new PacketByteBuf(PooledByteBufAllocator.DEFAULT.buffer());
buf.writeVarInt(0);
ctx.writeAndFlush(buf);
i'm doing this
(in a fabric mod)
not sure how correct this is
eh most of it is written by assuming things, but not reading tbh
sly crown
is 0 empty tho?
bleak current
wdym
ruby dust
If you want to get this error https://github.com/SpigotMC/BungeeCord/blob/master/protocol/src/main/java/net/md_5/bungee/protocol/Varint21FrameDecoder.java#L35 i think easy way is write 0 in VarInt encoder (if it works as i think)
bleak current
varint encoder, you mean that varint length prefix prepender?
i disabled compression for now, packet lenght = 1
🙃
length*
sly crown
yea i mean writing (varint) 0 costs 1 byte right...?
bleak current
yeah
which is 0x00
should be caught here
and length should be == 0
ghh must've something to do with my lack of netty skills
definitely doing something wrong and/or not understanding how pipeline works clearly
okay got it
stark bone
When doing ProxiedPlayer.connect() to a full server, is PlayerLoginEvent fired on the Sub server? I'd like to kick an AFK player when a user is attempting to join a full server
It's hard to test on your own
shut glacier
Could it be that waterfall preloads all classes of a plugin?
Or in other words, how can I use my own class loader?
trail plume
It doesn't preload them, you shoooould be able to use your own classloader without any issues beyond classloaders being classloaders
shut glacier
Weird. I’m explicitly loading my class with a custom class loader. And it still claims to have the bungee class loader
(A class separate from the plugin class)
bleak current
@shut glacier how are you loading the class?
shut glacier
Class.fromName(name, true, myClassLoader) @bleak current
dark perch
I think you should use this: https://docs.oracle.com/javase/7/docs/api/java/lang/ClassLoader.html#loadClass(java.lang.String)
I think
shut glacier
I can try
bleak current
dear lord
bungee's codebase contains too many shady arbitrary delays
e.g "wait 250ms before we kick the player"
yeah, fine - kicking the player at wrong time might crash it, but using a Thread.sleep(500) instead of proper state check smells really bad
or for example "wait 500ms after disconnecting all players" instead of checking whether all players are actually disconnected...
bleak current
Hello.
Is there any easier way then this to decide if a server is online or not?
getProxy().getServers().get(serverName).ping(new Callback<ServerPing>() {
@Override
public void done(ServerPing result, Throwable error) {
if (error != null) {
// Means that server is not responding : OFFLINE
// Store this, by example, in a Hashmap<Server,Boolean> serverStatus, false is OFFLINE and true ONLINE
return;
}
}
});
yes, wrap it into CompletableFuture
I would like to get it from a PaperMC plugin. Is there any way to check if a Waterfall connected server is online or not from my PaperMC plugin?
that's bit more complicated and involves either fucking with plugin channels a lot (slower, not very flexible and probably more unreliable because of programming errors)
or building a central server status daemon which checks whether servers are up (periodically/whatever mechanisms you can imagine) and provides that info via e.g simple http rest api
use your imagination
shut glacier
Great
Once I get my classloader to load my class I get a lovely java.lang.ClassCastException when trying to cast it to its bas class (which is loaded by the plugin classloader)
bleak current
move all the logic away into separate class
shut glacier
That would force me to call all methods by string names
bleak current
classloader hackery is a total time sink sadly
shut glacier
Yeah
All because Java doesn't have decent built in ways to inject libraries during runtime
Or enforce separation where necessary
warped loom
if i register a command on waterfall is it executable on every backend by default or not?
trail plume
It's a proxy, it proxies the commands and intercepts them
weary grove
In other words, it has no control over your backend
stark bone
How do I get the max players my proxy can hold?
ProxyServer.getInstance().getConfig().getPlayerLimit() is depricated and cant see anything else
trail plume
That's it, deprecation is basically just stupid there
stark bone
That's a shame. Really dislike using deprecated functions.
trail plume
Basically, the configuration class itself is deprecated because 'we might change the config'
Blame upstream, it's just typical blergh
stark bone
What do you think the likely hood of .getConfig() being yanked is?
trail plume
None
weary grove
Guava has the nice @UnstableApi annotation. I like that better.
trail plume
We already have a patch to fix up the javadocs from upstream, does IJ provide any hinting around that annotation?
I'm going to crash, but I have nothing against changing that stupid deprecated annotation to something more suitable, bonus points if the IDE offers some form of hint around it
weary grove
Yes, IJ does hint it
weary grove
@trail plume yeah, actually, the Guava annotation is @Beta, which is about the same thing
frank junco
so with Inventory instance, how does one go about checking if the inventory in InventoryClickEvent is the same inventory as the Inventory previously created?
i used to use .getName() but that has been removed
and Inventory does not appear to have the suggested getTitle()
dark perch
Does Waterfall/Travertine discard or stop plugin messages with unregistered channels?
weary grove
probably
actually, I just checked - upstream bungee doesn't do it, and I don't think WF does it either
bleak current
it doesn't yep
i haven't seen any downsides in that yet
besides bungee being not correct
cyan garnet
How do I get the player from the Prelogin event?
@EventHandler
public void onConnect(PreLoginEvent event) {
ProxiedPlayer player = ProxyServer.getInstance().getPlayer(event.getConnection().getUniqueId());
This isn't working
bleak current
you can't
player object does not exist at that point
what do you need player for there?
dark perch
The best you can do is store the latest know ip and compare that
cyan garnet
I need to check their permissions to see if they have a certain rank
and this is for an anti vpn plugin, want to check if they're a new player based on their group/rank and check their IP if it's a proxy,
ended up hooking into LuckPermsAPI and getting the user from there, that might work better, will try
bleak current
that's a way, yes
harsh aspen
Is possible listen the packet on BungeeCord/WaterFall?
bleak current
you have to inject your own packet handler into netty pipeline then
stark bone
any way to handle a player being kicked by "Server Full" and redirect them back to the lobby
I though ServerKickEvent but not so sure now
upper fiber
How can I serialize Enums?
sly crown
stark bone
Is there a way of getting the server the player will be connected to upon join?
Hopefully using PostLoginEvent
event.getPlayer().getReconnectServer() is null even though server is not default.
event.getPlayer().getServer() is also null
trail plume
Basically, no
That event is fired too early, before they've actually been connected or started initiating a connection to the server
stark bone
I did look at ServerConnectEvent as it seems to have a "JOIN_PROXY" reason. However, I can't seem to get that to fire at all.
I have a server queue system and if they queue exists, connect people to the lobby instead of the last server they are on
civic valve
@stark bone try ServerConnectEvent
you can use
if (event.getReason().equals(ServerConnectEvent.Reason.JOIN_PROXY)) {
event.setTarget(getProxy().getServerInfo("hubserver"));
}
Oh just saw you can't get it to fire
make sure you're putting an EventHandler annotation
stark bone
@civic valve jeez, note to self, put @EventHandler over events....
[21:48:51 INFO]: Event fired?
[21:48:51 INFO]: james090500 is trying to join creative
lmao
bleak current
ye
dark perch
Anyway to listen for Payload packet on Waterfall?
bleak current
payload packet?
bleak current
plugin message?
bleak current
you don't have to register them in bungee iirc
because bungee's channel registration is absolutely broken, half-assedly done and what so ever
so all plugin messages can be accessed via PluginMessageEvent
i think last time i touched it then it was like this
dark perch
I see, thanks
dark perch
Any ProxiedPlayer is an instance of UserConnection?
dark perch
This error is thrown when i send the ForgeConstants.FML_START_CLIENT_HANDSHAKE PluginMessage
It kind of get completed
But still kicks me
Seems like ForgeServerHandler is null
bleak current
is forge support enabled in config?
dark perch
There is any artifact like bungeecord-proxy for waterfall/travertine?
trail plume
Yes, same name format, they're just also not deployed
dusk crescent
how can i delete the players scoreboard? I get sometimes an error bc a team already exists
trail plume
You basically need to avoid having multiple servers using the same team names, it's basically a flaw in how bungees connection process works, best fix I've got is basically pending those packets for later in the connection phase, just not gotten to it
trail plume
Some levelof "good luck"
thorny holly
😐
trail plume
proxy logs are basically 100% irrelevant
thorny holly
figured
bleak current
you can allow waterfall to send the empty packet to the client
client simply ignores it
maybe waterfall does too if configured, can't remember exactly
oh wait... it's not configurable now :D
deft trail
is there a latest jar link? that doesnt specify a version number? like this but without the /109? https://papermc.io/api/v1/travertine/1.14/109/download
nevermind lol
dense ridge
What is the Repo vor Waterfall?
trail plume
same repo as paper
golden gulch
what is waterfall?
manic widget
golden gulch
ok
tired hedge
What’s the main new features of Waterfall comparing to Bungee?
remote condor
So i'm looking into automatically building my fork of waterfall with jenkins, what do you guys do to actually run the build? Like how does it know what maven tool to use if you execute the build script or will it always default to the maven installed on the system?
remote condor
Ya, after tinkering with it I managed to get it working with a very messy looking pipeline.
shut glacier
Ok, so I think I have a weird one for you
I'm working on a universal plugin (running on all major platforms) and to keep logging consitent I've opted for SLF4J
Now After some trial and error I managed to get a system working where if SLF4J is not present on the platform I load and inject it into the classpath before my plugin loads
So for example on Waterfall I use the built in version on Bungeecord I load and inject it myself
Now that works as should and is not the problem
The problem is that the logging looks awful on Bungeecord
That's how it looks on BungeeCord:
And this is how it should look and does look on Waterfall:
Does anyone have an idea on how I can get it to look at least similar?
stone fiber
Waterfall uses an updated version of the logger dependencies in bungee
and has explicit support for slf4j
shut glacier
I know
stone fiber
so the reason it doesnt look worth a damn in bungee is that, and the solution would be to go hope you can convince them it matters
shut glacier
I mean they use Java's logging system as far as I know
Now the thing I don't quite understand is why after adding slf4j-jdk14 the log is so weird
And how I can configure it to essentially just log to java.util.logging directly without stupid format code
If all it did was to match the log level, take over the logger name and the just print the message literally, it would be done
I could fix it by using a custom logging adapter
That sounds like a somewhat possible idea
golden gulch
how do i get rid of these warnings?
shell lagoon
Either properly check that your values are null; add @NotNull annotations if appropriate; or disable the inspection in your IJ Settings.
copper tree
If you have a command called test
And you give it aliases like t, testsmth
command.getName() will return test
But label can be t, testsmth or test
It's what players put after the /
bleak current
this channel is for waterfall dev
proper rampart
examples of using a custom log4j filter inside a bungee pl?
void lion
Hello ! Do you know how can i can filter my console log by words ? 🙂
slender grotto
weary grove
Invoking my magic looking glass... looks like the plugin isn't following all the semantics of BungeeCord AsyncEvent handling correctly.
The fix is to remove that particular plugin and file a bug report against it, or find a different plugin.
sly crown
nice advertising
ornate jasper
Wut
bleak current
The plugin is free and not obsucted if you want to decompile it.
might as well open source it?? xd
proper rampart
https://file.properties/paste/avociyojec.js <- Log4JFilter class
https://file.properties/paste/nuweyogemu.java <- LogFilterModule class
Also in my main class I have this: java ((Logger) LogManager.getRootLogger()).addFilter(new Log4JFilter(moduleManager));
(onEnable)
bleak current
so what is the issue in the end
proper rampart
There are empty messages rather then them just being cancelled.
I want to remove them, filter them out, not just replace them with nothing.
proper rampart
?
trail plume
take a look at console spam fix?
proper rampart
I want to implement a console filter in my plugin, not use external. Also isn't consolespamfix a paper plugin?
trail plume
There is a bungee version of it...
proper rampart
Didn't know that, didn't really searched for it. I tried to add the functionality to my plugin. Still looking for that.
I will check the github repo of csf. Thanks.
proper rampart
It's for bungee, not waterfall. 🙂
trail plume
To my understanding, should work the exactsame
proper rampart
If I'm not wrong, waterfall uses log4j, not regular log. CSFB doesn't have the log4j support.
trail plume
Have you tested it?
proper rampart
It works but it doesn't filter. I'm using a custom waterfall fork (from a dev that added some antibot functions, premium) so it may not work because of that.
I'll ask him for it too.
proper rampart
Strange.
I'll try with a regular waterfall build.
Enabled Waterfall version git:Waterfall-Bootstrap:1.14-SNAPSHOT:cccb337:296
¯_(ツ)_/¯
proper rampart
Nope.
trail plume
Not even CSFB handles the UpstreamBridge one properly
proper rampart
For reference my plugin filters them out properly but sends spams messages.
Oh it seems that I've fixed it.
private Filter.Result checkMessage(String message) {
for (String string : moduleManager.getLogFilterModule().getFilter()) {
if (message.contains(string)) {
return Result.DENY;
} else if (message.isEmpty()) {
return Result.DENY;
}
}
return Filter.Result.NEUTRAL;
}
I used a for loop and checked if the message is empty.
works as intented
Thanks cat!
knotty delta
What causes texturepacks to load twice when using bungeecord/warterfall?
Like when changing between two servers with different texturepacks for example
trail plume
You'd need to send an empty texture pack to load a new one
knotty delta
No I mean I understand you'd need to do that, but it's like it does it no matter what even if it already had an empty texture pack it loads the same new texture pack twice
I can make a video to show you if it's helpful to understand what I am asking
trail plume
I mean, it'd only load it twice if it was told to load it twice
knotty delta
it only happens when using bungee/waterfall
like when I coded it into a spigot plugin to load the pack it did it once when running spigot alone, but through bungeecord it loaded it twice even when just loading the empty pack on the servers without packs
trail plume
You don't still have the plugin on the servers, do you?
knotty delta
yes but I removed the texturepack loading from my plugin and tried using a bungeecord plugin
it has the same behavior, it loads twice still
proper rampart
What is the maven dependency for net.md_5.bungee.command.ConsoleCommandSender ?
bleak current
waterfall-api
bleak current
...bungee.api.command.ConsoleCommandSender ?
never mind, it's an impl
it's in waterfall-proxy artifact
but it's not deployed
if you want to check if sender is console, use if (!(sender instanceof ProxiedPlayer) && sender.getName().equals("CONSOLE"))
real cave
ProxyServer.getInstance().getConsole()
Returns console command sender
Now you can just do ==
bleak current
ah yes, that's a way too
dark perch
Does the delete scoreboard team packet helps with the issue of players being kicked due duplicated team name?
old mason
is there a module to add a /join command to join servers insted of /server? if so could you direct me to it
deft tangle
Hello, is there any way to get the ProxiedPlayer instance from a ServerConnection?
lean gobletBOT
(DiscordBot) https://github.com/SpigotMC/BungeeCord/blob/master/proxy/src/main/java/net/md_5/bungee/UserConnection.java -- BungeeCord/UserConnection.java at master · SpigotMC ...: "BungeeCord, the 6th in a generation of server portal suites. Efficiently proxies and maintains connections and transport between multiple Minecraft..."
bleak current
or not
iterate over players, check if player's server connection is the said connection
don't know better solution from top of the head right now
trail plume
to scream at people using outdated versions of java
Just bump it to 1.7 or something
stone fiber
Yeah we took that java 6 warning out of paperclip for that reason
Most people kinda jumped off that ship when Mojang made them
Not sure where the proxies are, might be feasible to consider the same but eh ¯_(ツ)_/¯
nocturne zinc
@night grotto @trail plume @lofty ridge i have a Problem a Bad guy crashes bungee Servers easyly by using a indexoutofbounds exception at net.md5.bungee.protocol.Minecraftdecoder at a Check in readablebytes are 0 tomorrow i Could Senf you a log today im net longer at pc and ihrer Servers like hypixel was crashed by him too please create a fix and publish it Soon bye Max
tribal jacinth
yikes
lofty ridge
that's nice
trail plume
Literally already an issue on the tracker for that too
tribal jacinth
did you kick him or did he just leave
trail plume
I kicked
tribal jacinth
bleak current
i implemented automatic upstream commit merging into my concourse pipeline for a while ago...
let's see if it actually works
||here's the link for the curious: https://ci.mikroskeem.eu/teams/mikroskeemsrealm/pipelines/mikrocord/jobs/try-upstream-merge/builds/2||
lel it did naht
sharp onyx
any website u know i can learn these stuff
trail plume
.rekt
lean gobletBOT
☑ Tyrannosaurus Rekt
trail plume
Dude, pick a channel and stick to it
sharp onyx
ye mb
bleak current
so ye that did not kick in
wait
it's my script after all
but that means i'll mod the shit out of it
trail plume
Tryna blame me u fuq
bleak current
cat bad
sharp onyx
oh wait i thought i was at paper channels mb lmao
trail plume
concourse, urgh
Now I got a new toy on the list to play with
I do wanna get to grips with pipelines on jenkins too as they look interesting; I just dun has the brain
bleak current
concourse is a bit bitch to set up btw
so just warning
also if you happen to use ZFS then...
tho wut
set -e is missing from the main script
nvm
ah build script needs it too...
trail plume
bleak current
better
otherwise if it succeeded, it should've pushed a change with rebuilt patches & updated submodule 😄
bleak current
-> #paper-help
trail plume
looks at the channel name
bleak current
why bother reading
bleak current
new upstream commit
me being patient for automatic merge to succeed
failed
aight, looking what's up:
doing same thing locally
it works
wut in tarnation
trail plume
.rekt
lean gobletBOT
☑ md_rekt
ornate jasper
well that rekt was rather convenient
bleak current
.rekt
lean gobletBOT
☑ Better Dead Than REKT
bleak current
weary grove
Hehe, BungeeCord development is ramping up a bit again
bleak current
yep
trail plume
Revert "#2714: Remove unnecessary throws in ServerConnector"
This reverts commit 74a6aa3.
Completely breaks Bungee
ripperino
idle lance
where is bungee's changelogs/commits?
trail plume
idle lance
ah ty
civic valve
Glad to see md_5 testing pull requests
burnt knoll
does anyone knows a good way to sync data through all the servers on the proxy?
trail plume
Because plugin messages are sent through a users connection
The proxy doesn't maintain its own connection to the server for sending messages, that wouldn't be supported by the mc protocol
If you rely on those messages, better off using a read message broker vs the hackjob that is plugin messaging
*real
burnt knoll
well I better start learning sockets
trail plume
better off looking into a message broker vs sockets, especially if you want your plugin to be used in scalable environments
frosty grove
https://pastebin.com/XdwJhW5M I've tried deleting my scoreboard files, any help?
frosty grove
? I'm pretty sure it's on by default and I never disable player collision?
remote condor
Do you use protocol hacks like ProtocolSupport or ViaVersion
frosty grove
nope
remote condor
Idk, those are the most usual culprits, either way it is just a warning with no actual ill side-effects
frosty grove
whenever it shows in waterfall everyone gets kicked
remote condor
What plugins do you run on the bungee/servers
Something has to be causing that.
The logs you provided don't have the actual scoreboard name for anyone to point out a plugin
frosty grove
yeah
because there isn't
in the console it's just Team and a space
that's my plugin list
any help?
remote condor
I'm going to guess its a factions scoreboard, but I havent ran factions for years so idk.
tribal jacinth
factions
bleak current
Hello.
I have an ArrayList in my Waterfall plugin and I would like to add items to this list from the web with POST request.
Example: "www.myserverip/plugin?name=asd" will add "asd" into the ArrayList.
Is this solvable?
Note: The "www.myserverip/plugin?name=asd" link will be not public only allowed for 1 remote IP.
Is this a good solution or I need to use a database and PHP etc.?
bleak current
wut
so you want to pull up a webserver on waterfall and get said POST requests?
... why shouldn't that be solvable?
everything can be solvable if you can write code :o
tribal jacinth
ancestor simulation incoming
bleak current
trail plume
Didn't notice it added the iml file, and I coulda worn I removed the DS_Store file 
In dedication to the fact that I fucking hate myself, have fun: https://keybase.pub/electronicboy/builds/travertine/Travertine-e115.jar
Unsupported, doesn't support entity metadata remapping either as I'm too lazy/dead and it's going to be removed when bungee updates anyways
sly crown
hating yourself isn't healthy cat
you should find something better to unleash your hate on, like people asking "paper 1.15 when"
shell lagoon
I'll help cat release his internal anger: "paper 1.15 when"
trail plume
an fix'd a derp as I forgot to mess with the respawn packet
drifting crest
How much Traveltine 1.15 can broke?
random comet
hmmm whats entity metadata remapping then
trail plume
entity metadata remapping is basically rewriting the entity metadata so that entity IDs for players are written properly as bungee doesn't send the client its new entity ID when it jumps servers
Waterfall has an optional mode, which is enforced for the 1.15 protocol, which will send the client its correct entity ID, but basically is "less transparent" to the client
random comet
oh is that why players would still be flying with elytra if they were when switching servers
or whats the side effects
trail plume
The mech without the rewriting would actually sort that one out by default afaik
(to reset the entity ID, we basically have to tell the client to do a bit more of a through state cleanup)
Only real side-effect is that plugins aren't always happy with the mechanism, pretty sure one of the tab list plugins gets a bit upset with some ordering with it
deft trail
cat, tried out that travertine build with 1.15 support. i can connect to my hub server just fine, but whenever i switch servers it kicks me for this? or would this be viaversion territory?
trail plume
You'd need to work out where that's being thrown from, it's basically upset that a packet is shorter than it's expecting
trail plume
that one just states that the connection was closed improprely by something else
random comet
@lilac crystal fix ForceResourcePacks please 
https://pastebin.com/raw/3tEZZZkc
deft trail
tried it on a blank 1.8 server with viaversion (yes ik 1.8, but its the only server i have atm that i could wipe, rest have stuff on it i need), this is saying the same thing and doesnt really help right? https://pastebin.com/U0a15ywL
trail plume
No, broken pipe is a broken pipe
All that means is that something rekt the connection
deft trail
ugh
does cross server switching work for you on 1.15? if you dont have a test environment or if you're not up for it, nvm.
trail plume
Yup
[19:44:56 INFO]: [electronicboy|/192.168.194.1:59681] <-> ServerConnector [valaria] has connected
[21:17:56 INFO]: [electronicboy|/192.168.194.1:59681] <-> ServerConnector [lobby] has connected
[21:17:59 INFO]: [/192.168.194.1:59681|electronicboy] <-> DownstreamBridge <-> [valaria] has disconnected
on the backend server, yes
deft trail
hmm
im using the plugin on the server
clean 1.14 server, only plugin is viaversion, same kick/error. ima try switching to 1.15 server with no viaversion
Exception Connecting:QuietException: Unexpected packet received during server login process!
trail plume
If that's a vanilla server, disable ip forwarding
random comet
I have it working on my server 
deft trail
viaversion on bungee does the same thing
Do you use BungeeCord? Ensure you have updated it first, you can use the plugin on all your backend servers OR BungeeCord. Compatibility is the best when on your backend servers.
random comet
seems backward to me
deft trail
ye haha
random comet
yea i put it on spigot servers
random comet
U can put via on either
sly crown
so why is it backward
they are saying it's better to put it on the spigot server
and you said you put it on the spigot server
but you also say it seems backward
????
random comet
it seems backward to have viaversion on bungee when bungee natively supports various versions
bleak current
via on bungee makes sense only when uh
you want snapshot versions...?
or whatever
deft trail
heyyyy cat, ./waterfall up 😃
bleak current
trail plume
If he's pushed 1.15 for bungee, that shoooould mean that he's planning to drop spigot tonight soon
trail plume
31 mins ago according to jenkins
civic valve
time to turn on builddata notis
final lagoon
can we download 1.15 ?
civic valve
no
random comet
theres only bungee i think
bleak current
"reeee paper succccc no 1.15"
final lagoon
dont be like that lol
civic valve
inb4 "guys spigot 1.15 out why no paper 1.15 jus d0 ./paper up??!"
final lagoon
paper up from console says unknown command thanks
bleak current
mfw
final lagoon
im running a windows server if that helps, my options are heap entity reload version debug chunkinfo
bleak current
... -> #paper-help
trail plume
They're referring to the paper scripts used by devs; Basically, just wait, spigot hasn't released 1.15 yet
final lagoon
ah ok thanks, Spigot has released 1.15 I will wait. thanks all
remote condor
So when's md5 removing 1.8 support :^)
random comet
@trail plume I get this with travertine off build server https://gyazo.com/9494420f7104371dc2419b92d7488118
sly crown
set ur fukin game version
random comet
?
trail plume
Worked here fine
random comet
I have game_version: ''
is it supposed to be something else
weird i guess i downloaded the wrong version
maybe i pressed the back button on my mouse and latest was still old link thats only explanation I have 😓 sorry
random comet
would be neat if you could make that always show, and change the color of it
trail plume
There are plugins for that
Pretty sure they can do the color too; Making it show is a case of lying about the protocol version, however
random comet
o interestin
deft trail
i updated from that jar (https://discordapp.com/channels/289587909051416579/555462232209621003/654024941284491293) and im no longer getting kicked when switching servers, really weird.
high lily
what is travertine
deft trail
what are you even saying lol. cat made a travertine jar with 1.15 support before bungee released theirs. all he mentioned was that entity metadata remapping wasnt supported. he never said it was broken, and it definitely wasnt old when he released it. i thought hey why not, someone might as well try it out. so I threw it on and tested it out while i waited for bungee to release. i could connect but i couldnt switch servers without getting kicked with the error i posted about above.
all im doing is posting a followup to my issue.
bleak current
Hello, how can I cancel ProxyPingEvent? event.setResponse(null); works but always drop a nullpoint exception.
bleak current
you can't set it to null
bleak current
How can I cancel it? So player see "Can't reach server".
.g bungeecord proxypingevent
lean gobletBOT
(DiscordBot) https://github.com/SpigotMC/BungeeCord/blob/master/api/src/main/java/net/md_5/bungee/api/event/ProxyPingEvent.java -- BungeeCord/ProxyPingEvent.java at master · SpigotMC ...: "BungeeCord, the 6th in a generation of server portal suites. Efficiently proxies and maintains connections and transport between multiple Minecraft..."
bleak current
.g bungeecord initialhandler
lean gobletBOT
(DiscordBot) https://github.com/SpigotMC/BungeeCord/blob/master/proxy/src/main/java/net/md_5/bungee/connection/InitialHandler.java -- BungeeCord/InitialHandler.java at master ·...: "BungeeCord, the 6th in a generation of server portal suites. Efficiently proxies and maintains connections and transport between multiple Minecraft..."
bleak current
use reflection to get ch field from PendingConnection instance
then call getHandle() and you have an access to netty Channel
further steps should be obvious
ooor you could use PendingConnection#disconnect()
misty elk
Hello, i want to maintaining the proxy for a long, so i am trying to plugin segmentation, is it good? it could be ~20 plugins
bleak current
yes
misty elk
i just going to split the features from plugins
sorry for being dumb at english, didint learn as well
misty elk
is it cause performance problem? and it affects a lot?
frigid brook
Can someone tell me how to detect if a player holding a carrot?
trail plume
- Wrong channel
bleak current
it's kinda shitty belief that n amount of plugins causes performance issue
rather, multiple plugins doing slow things causes performance problems
sly crown
in fact, one plugin doing slow things also causes performance problems
tidal rover
Back again, can someone help with this Maven compile error? https://i.imgur.com/p6ngKGc.png
tidal rover
Okay thank you, I'm looking through Maven's settings in IntelliJ, I've set "maven.compiler.source = 8". Though, I feel this is wrong as it's still failing.
Though, with a completely different error this time.
lean gobletBOT
(DiscordBot) https://maven.apache.org/plugins/maven-compiler-plugin/examples/set-compiler-source-and-target.html -- Apache Maven Compiler Plugin – Setting the -source and...: "Apr 28, 2019 ... Setting the -source and -target of the Java Compiler. Sometimes when you may need to compile a certain project to a different..."
tidal rover
Okay, thanks again. 🙂 I'll take a look.
That worked. Damn that was simple. Thank you.
sly crown
what's the question
sly crown
are you asking for waterfall's maven location
golden gulch
uh
wrong channel
paper-dev?
also i just don't get what i am meant to put in there
what group id
what group
golden gulch
my plugin
sly crown
you just gave a screenshot of a random dialog box
golden gulch
yeah
sly crown
unclear what you are trying to do
golden gulch
i'm making a new project
sly crown
groupid for your plugin is whatever you make it
you should probably google up a guide on using maven
golden gulch
does it just mean like an identifier?
sly crown
and not use the waterfall channel
cerulean kettle
I think that a option to disable the exceptions should be added to the config, i just disabled exceptions and a little more stuff and all crashes because of invalid packet spam has gotten fixed.
Took me 5 minutes to fix a really common exploit that is being really abused this days.
trail plume
Submit a PR...
Cleaning up the exception handling in waterfall is on my todo list, just 1.15 and my health in general is somewhat limiting
bleak current
I'm lost in how do I build waterfall? I get this error:
[ERROR] [ERROR] Some problems were encountered while processing the POMs:
[ERROR] Child module /root/waterfallsource/Waterfall-master/Waterfall-Proxy of /root/waterfallsource/Waterfall-master/pom.xml does not exist @
@
[ERROR] The build could not read 1 project -> [Help 1]
[ERROR]
[ERROR] The project io.github.waterfallmc:waterfall-super:dev-SNAPSHOT (/root/waterfallsource/Waterfall-master/pom.xml) has 1 error
[ERROR] Child module /root/waterfallsource/Waterfall-master/Waterfall-Proxy of /root/waterfallsource/Waterfall-master/pom.xml does not exist
[ERROR]
[ERROR] To see the full stack trace of the errors, re-run Maven with the -e switch.
[ERROR] Re-run Maven using the -X switch to enable full debug logging.
[ERROR]
[ERROR] For more information about the errors and possible solutions, please read the following articles:
[ERROR] [Help 1] http://cwiki.apache.org/confluence/display/MAVEN/ProjectBuildingException
but I can't find waterfall-proxy for download anywhere
trail plume
use the scripts
bleak current
which?
trail plume
Well, *the script
bleak current
I'm confused, on the github page all it says is:
How To (Compiling From Source)
To compile Waterfall, you need JDK8, git, bash, maven, and an internet connection.
Clone this repo, run ./waterfall b from bash, get jar from Waterfall-Proxy/bootstrap/target/
never mentions use of any scripts, so I really have no idea
trail plume
What do you think "waterfall" is?
bleak current
proxy server?
civic valve
smh
trail plume
That waterfall file is literally a script
bleak current
ohh
trail plume
The instructions right there literally tell you exactly what to do
bleak current
I still get the same error
trail plume
would need to see the full log
bleak current
[ERROR] Child module /root/waterfallsource/Waterfall-master/Waterfall-Proxy of /root/waterfallsource/Waterfall-master/pom.xml does not exist @
at org.apache.maven.project.DefaultProjectBuilder.build(DefaultProjectBuilder.java:422)
at org.apache.maven.graph.DefaultGraphBuilder.collectProjects(DefaultGraphBuilder.java:419)
at org.apache.maven.graph.DefaultGraphBuilder.getProjectsForMavenReactor(DefaultGraphBuilder.java:410)
at org.apache.maven.graph.DefaultGraphBuilder.build(DefaultGraphBuilder.java:83)
at org.apache.maven.DefaultMaven.buildGraph(DefaultMaven.java:491)
at org.apache.maven.DefaultMaven.doExecute(DefaultMaven.java:219)
at org.apache.maven.DefaultMaven.doExecute(DefaultMaven.java:193)
at org.apache.maven.DefaultMaven.execute(DefaultMaven.java:106)
at org.apache.maven.cli.MavenCli.execute(MavenCli.java:863)
at org.apache.maven.cli.MavenCli.doMain(MavenCli.java:288)
at org.apache.maven.cli.MavenCli.main(MavenCli.java:199)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.codehaus.plexus.classworlds.launcher.Launcher.launchEnhanced(Launcher.java:289)
at org.codehaus.plexus.classworlds.launcher.Launcher.launch(Launcher.java:229)
at org.codehaus.plexus.classworlds.launcher.Launcher.mainWithExitCode(Launcher.java:415)
at org.codehaus.plexus.classworlds.launcher.Launcher.main(Launcher.java:356)
[ERROR]
linux
sorry I'm new to building stuff, so I'm not sure what exactly "full log" is xD
trail plume
I'm ill and really do not have the mental capacity to deal with this right now
But like
Wtf do you think the word "full" means
bleak current
well good luck, I will find other way maybe
bleak current
this is full log I hope, but me going off so may only reply tomorrow
https://pastebin.com/raw/s3LHQMUJ
trail plume
root@limework:~/waterfallsource/Waterfall-master# sh waterfall b -e
fatal: Not a git repository (or any of the parent directories): .git
Literally the first step
clone this repo
bleak current
speaks about negative iq
still calls paper as paperspigot
... can you atleast try? :D
either way
http perhaps?
bleak current
i've used redis/rabbitmq
depending on a thing
if i need rpc-like comms then grpc or http (json rest api) does the trick
cerulean kettle
@trail plume Okay, thanks for considering it. I wanted to add a config value but its the first time i fork travertine and i dont really know how to do it.
plus the handshake event is being called even if the protocol is not 1 or 2 and there where some more stuff running that souldnt be running if the packet is invalid.
@bleak current It is a exploit actually, as you're exploiting a vulnerability to generate unwanted consequences. (in this case generating exceptions like hell by spamming invalid packets)
tribal jacinth
ping some more staff while you're at it
bleak current
unwanted consequences such as?
verbose logging?
have you tried reporting to openssh devs that you're getting bombarded by spambots, who generate 3-4 log lines per session, that it's an exploit? :p
tribal jacinth
several terabytes of logged stacktraces later
bleak current
I always thought that cloning this repo means downloading it manually from github (recommendation: include that you need to use "git" command to clone it instead of manual download)
maybe let's create a youtube tutorial how to build waterfall
and link it into the readme
🙄
bleak current
eh wouldn't use it, video tutorials are extremely boring xD
so it's easier to learn the difference between (git) cloning and downloading
trail plume
it literally uses the word "clone", not "download and extract the zip file" or anything like that
bleak current
Hello. Do you know something about "ping attacks" when bots not connect to the server only ping them? It works like when I refresh the server list? How can I protect against it? I have to cancel the ProxyPingEvent (disconnect the pending connection or set different result)? Or use the server firewall is better?
your potential aid is to set up a firewall rule to rate limit the connections
deft trail
I wrote some code that grabs the servers from a database table and uses
getProxy().getConfig().addServer(serverInfo); thru the result set. it works all fine and dandy, but when bungee starts up, it cant find the servers set in priorities and errors since the server retrieval query takes place in the bungee plugin. also errors with forced_hosts too. any idea on how to approach fixing that?
deft trail
onLoad(), exactly what im looking for. thanks!
misty elk
Hello, how can i change server name for waterfall?
(screenshot was from paper without waterfall because didnt have time)
bleak current
use a plugin
misty elk
yes this is development channel and i looked out javadocs and i dont know what to do because bad at english
i mean im creating waterfall plugin
lean gobletBOT
(DiscordBot) https://www.spigotmc.org/resources/f3name-edit-your-server-brand-in-debug-screen.58997/ -- F3Name - edit your server brand in debug screen | SpigotMC...: "With this plugin, you can easily edit server brand from the default ("Spigot", or very long Bungee version) to whatever you want. Show your ..."
bleak current
there you go
misty elk
https://github.com/prettydude/F3Name/blob/master/src/main/java/ua/coolboy/f3name/bungee/F3NameBungee.java#L296
thanks figured out
tidal rover
Is there anyway to 'dev' live? Like can I hold a server open and run methods to see what the server returns?
Idk if I'm wording the question correctly.
I find it's quite slow writing something (that doesn't work, because I'm very new to Java programming), compiling it, uploading it and restarting the server. There must be a faster way, right?
shell lagoon
You can setup a remote debug session to your server and then insert breakpoints in the code and the server will then just halt, so you can do your debugging.
tidal rover
Okay, thank you. I'll take a look into how to do this.
boreal atlas
There's any build compatible with 1.15.1?
shell lagoon
No
civic valve
quartz turret
What do you mean by "git not to break"?
You want to still be able to do "git pull"?
signal mica
What's the safe way of adding a server to the Waterfall Proxy with a plugin?
trail plume
collection returned by getServers is mutable
deft trail
signal mica
Thanks. Is this already in maven? I don't find the method in my IDE.
deft trail
are you using waterfall or bungee in your maven
signal mica
waterfall
deft trail
do you also have bungee in there too
signal mica
No. My proxy only uses waterfall and my plugin only has the waterfallmc dependency.
signal mica
I am using this in maven:
<dependency>
<groupId>io.github.waterfallmc</groupId>
<artifactId>waterfall-api</artifactId>
<version>1.15-20191221.065800-7</version>
<scope>provided</scope>
</dependency>
p sure its just waterfall and ur version number should be 1.15
trail plume
waterfall-api is correct, but yea, should be 1.15 too
signal mica
I pulled this weired version out of nexus. 1.15-SNAPSHOT also doesn't work 😦
trail plume
mvn dependency:tree
signal mica
trail plume
What method are you looking for exactly?
signal mica
ProxyServer#addServer()
trail plume
That would be why
deft trail
lol
trail plume
deft trail
ProxyServer.getInstance().getConfig().addServer()
signal mica
Okay wow. It's officially too late for me...
I have already been decompiling my waterfall version to search for errors...
Thanks for the help.
pale gate
Hey, anyone knows that? I'm using travertine: https://pastebin.com/9b69QxkH it just crashed or something like that
sly crown
DoS?
cerulean kettle
Hey, im trying to compile waterfall (travertine) but when importing objects from api in a protocol class it gives a error. (I dont really know how to explain correctly, but heres a pic of what i mean) https://i.imgur.com/NXCKtOJ.png
I hope someone close to the source knows how to fix this. Thanks.
cerulean kettle
Oh okay, i wanted to use ProxyServer.getInstance() to access some stuff there. You know if theres another way to do so?
trail plume
No, you need to devise another way to pass the stuff you want over
cerulean kettle
Okay, thanks :)
bleak current
you can do quite hackish thing with suppliers
e.g if you want a boolean flag from config or something
then you can abuse a static field where you'll set a booleansupplier instance
lombok makes it tad prettier as well lol
deft trail
im in MinecraftDecoder class trying to make some edits, I want to grab the config with BungeeCord.getInstance().getConfig()
however it wont let me autoimport BungeeCord, i added
import net.md_5.bungee.BungeeCord; to the top but when i go to build it, it cant find that either. how do i grab the config from net.md_5.bungee.protocol.MinecraftDecoder?
trail plume
That module doesn't import the API
deft trail
ah, so its not possible to grab the config from there?
trail plume
no
deft trail
well fuck, so much for this pr lol
bleak current
what pr
trail plume
Permissions...
Oh, doesn't have a perm, lol
..
My comment was more, "if they don't have perms to use the command, they won't see it"
Then I noted that that command doesn't have a perm node, and linked a pretty plugin for you...
It's a waterfall plugin...
bleak current
it is, everything is stuck behind your imagination
molten jackal
How should I go about sending a resource pack over waterfall (programmatically)? I can't find an event that is called once the player is loaded into the server (similar to Spigot's PlayerJoinEvent). Sending the resource pack when the player connects or starts to connect is just not the best of ideas I'd imagine. I could hypothetically do it through Spigot, but that would make the player have to update their resource pack every time they join a new server. I want to send the pack once they join the bungee, and not have to download it again until they join again.
I could also hypothetically send a plugin message from the bungee to the server they are connecting to at the first join and then have the server send it.
Tag me if you have a response.
trail plume
phoenix has a plugin which does just that
You basically need to register the thing yourself, but, pretty sure that the client is smart and won't reapply the pack, assuming that you set up the right info for it
molten jackal
Is there source code available?
I’m looking to use only like ProtocolLib and ViaVersion as my non-swedzian plugins
It honestly seems like easy enough of a system to implement, just send a packet from bungee to the player when they join
trail plume
Use the plugin messaging channel then
molten jackal
Like what I said?
I’m going to try this later https://www.spigotmc.org/wiki/send-title-to-player-packets/
Just send the resource pack packet instead.
trail plume
There is literally already API to do everything you're talking about
There is an API to send messages between the proxy and the server
molten jackal
Is there some sort of bungee event that is called when fully join?
molten jackal
I don’t want to do it between server and proxy, would prefer to do it all on the proxy side
stark bone
not sure if i'm going crazy here.... I'm getting a lovely error:
[21:12:18 ERROR]: [Ping Handler] -> creative - exception processing exception
java.lang.NullPointerException: null
at net.capecraft.bungee.commands.AfkCommand$1.done(AfkCommand.java:50) ~[?:?]
at net.capecraft.bungee.commands.AfkCommand$1.done(AfkCommand.java:1) ~[?:?]
at net.md_5.bungee.connection.PingHandler.exception(PingHandler.java:51) ~[server.jar:git:Waterfall-Bootstrap:1.15-SNAPSHOT:4ecb3f3:309]
at net.md_5.bungee.netty.HandlerBoss.exceptionCaught(HandlerBoss.java:167) ~[server.jar:git:Waterfall-Bootstrap:1.15-SNAPSHOT:4ecb3f3:309]
When doing
player.getServer().getInfo().ping(new Callback<ServerPing>() {
@Override
public void done(ServerPing result, Throwable error) {
//Checks if server is full else do AFK check
if(result.getPlayers().getOnline() == result.getPlayers().getMax()) {
The null point is on the "if" statement. Not sure why it's null though
Now this was working. The only change is I have moved server hosts and update to 1.15.1 from 1.14.4
My original though is the waterfall can't contact the server, but I can connect just fine
trail plume
something you're calling on is null
result or getPlayers
Oh, result is gonna be null
Check for the throwable first
stark bone
The throwable shows me the same error
Oh the result is not null
the get Players is null
trail plume
The throwable will be null of it was successful
stark bone
[21:16:48 INFO]: ServerPing(version=ServerPing.Protocol(name=Paper 1.15.1, protocol=575), players=ServerPing.Players(max=4, online=2, sample=[ServerPing.PlayerInfo(name=james090500, uniqueId=ba4161c0-3a42-496c-8ae0-7d13372f3371), ServerPing.PlayerInfo(name=mov51, uniqueId=bf8b08a5-714c-4667-8f49-efce56cb7dc5)]), description=A Minecraft Server, modinfo=ServerPing.ModInfo(type=FML, modList=[]))
[21:16:48 INFO]: ServerPing.Players(max=4, online=2, sample=[ServerPing.PlayerInfo(name=james090500, uniqueId=ba4161c0-3a42-496c-8ae0-7d13372f3371), ServerPing.PlayerInfo(name=mov51, uniqueId=bf8b08a5-714c-4667-8f49-efce56cb7dc5)])
[21:16:48 INFO]: null
System.out.println(result);
System.out.println(result.getPlayers());
oops
updated the messy response
So result is not null but result.getPlayers is null
trail plume
You have players in that output though?
stark bone
It may be running twice, for some reason.
let me double check this
Ok...
So,
So the calback runs twice
but the main method runs once
and then the result is null on the second run
[21:22:14 INFO]: Main Method Running!
[21:22:14 INFO]: Call back Running!
[21:22:14 INFO]: ServerPing resultxxxxx (removed to avoid spam)
[21:22:14 INFO]: Call back Running!
[21:22:14 INFO]: null
Why it is running twice is a mystery to me
Ok, I got it, so some how, its caused by LuckPerms. Actually printing the error System.out.println(error); rather than doing a printStacktrace shows me this. ```[21:32:51 INFO]: java.lang.NoClassDefFoundError: Could not initialize class net.capecraft.bungee.helpers.LuckPermsHelper
Due to upgrading 5.0 and needing to update the API calls
trail plume
How did one even o.O
stark bone
I mean, it's strange that the callback ran twice after the error and that one worked and the other errored
very confusing
trail plume
What my guess is is that that plugin is hookinh in somewhere and just being 10/10 janky somehow
stark bone
Thank you for your help anyway. The joys of plugin development
proper rampart
because people got used to using ipwl and have probably never even heard of bungeeguard 🙂
celest perch
If you know how to correctly setup your proxy server
You don’t need either
Firewalls, IP tables, and NYS
chilly coral
hello
i see you added ConnectionInitEvent to waterfall, i would suggest an improvement
when you use Waterfall behind HAProxy, the event become quite useless as the ip the event see is the ip of the HAProxy server
i case of HAProxy, i think this Event should also be triggered after the HAProxy sent his packet with the real player IP
to differenciate these two differents event trigger, a boolean could be added into the event to tell if the event comes "before HAProxy" or "After HAProxy"
Thank you all for your work on this fork 🙂
trail plume
That would be a different event given that it occurs later on in the process, after the actual init
chilly coral
you're right
i think a warning should also be added into the ConnectionInitEvent javadoc to tell that if a loadbalancer is configured, the IP in the event is from the loadbalancer
lean gobletBOT
I mean the event doesn't seem to make any guarantee that this is the actual player client connection
chilly coral
why ?
lean gobletBOT
probably because it's called before it is even checked if whatever is connecting is a notchian client
trail plume
The event only garauntees that it's a connection being initiated
lean gobletBOT
basically: if you need player information use one of the player events
chilly coral
yes, that's why i suggested an improvement or a new event for the proxy-protocol case
trail plume
Getting the IP from haproxy would be later in the connection phase, meaning that it couldn't be covered under the existing event
Create an issue on the tracker
I'll have to have a look when my head isn't killing me or something
lean gobletBOT
how does waterfall even add support for hyproxy?
*haproxy
or is that something bungee supports by default?
chilly coral
it already has
snow bolt
bungee supports it
lean gobletBOT
ah, I see
snow bolt
it just transmits haproxy's protocol information
trail plume
Bungee supports it, it basically adds a decoder that understands packets from haproxy and just faux's the IP there
lean gobletBOT
so wait, can people just fake their ip that way? lol
snow bolt
it has to be enabled in the config
trail plume
If you've got haproxy enabled, yes
trail plume
ye
lean gobletBOT
hm, the decoder is from netty directly
chilly coral
i will submit a ticket on the tracker. With a plugin that throttle connections, this kind of modification could help with the "error flood hack".
if you properly configure your iptables between your waterfall and your haproxy, there's no reason a player could fake his IP.
just like the IPTables beween your waterfall and your mc servers
lean gobletBOT
I just thought it might be enabled by default
lean gobletBOT
(DiscordBot) https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt -- PROXY protocol: "2017/03/10 Willy Tarreau HAProxy Technologies The PROXY protocol Versions 1 & 2 Abstract The PROXY protocol provides a convenient way to safely ..."
fading blaze
hoping with fingers crossed that this is an issue you are working on and im not uniquely affected
trail plume
yes
fading blaze
awesome, is there any suggestions that you reccommend to mitigate in the meantime?
(sorry didnt mean to delete previous comment was clicking copy)
trail plume
That fix stops the crashing from that specific issue
Just, you'll wanna have something like fail2ban to deal with the actual connections themselves
fading blaze
i have limited connections in my iptables to no more than 3 per ip, so that is mitigating it somewhat. but still pushing cpus to high levels
fading blaze
Latest build virtually mitigated the impact. thansk guys
random comet
with your last update to travertine it breaks @lilac crystal ForceResourcePack plugin, idk if he needs to update or there is something wrong with travertine, but players just kicked: https://hastebin.com/oqohabimev.rb
@trail plume (sry if i shouldnt tag)
snow bolt
the latest two updates are upstream patches
specifically the ones that speeds up common exceptions, which seems to be related to this
sly crown
or you could file a proper issue instead of pingimg everyone
trail plume
The paste doesn't even work...
sly crown
hastebin 🚮
snow bolt
sly crown
also trash tbh
snow bolt
eh atleast it works, saves pastes for quite some time, doesn't go down often
random comet
Normally hastebins fine idk why it deleted so fast
sly crown
in my experience it's "normally" 🚮
stone fiber
^
I dont even recommend it anymore, sometimes files last a few days, other times a week or two, but nothings ever still there when I want to look at it later
paste.gg and gist, hell even pastebin
snow bolt
it's supposed to delete files after 30 days
sly crown
regardless of what it's supposed to do, that's clearly not the case in reality
as was just demonstrated
and given that we've had these experiences many times previously, this also isn't new
so as suggested, just refrain from using it
random comet
update to 120, everyone gets kicked with that, downgrade to 118 its all fine
i can submit a issue on github but i dont have any more info
trail plume
o.O
random comet
I use this but idk if it needs to be updated or travertines fault https://www.spigotmc.org/resources/force-resourcepacks.10499/
trail plume
Basically, that doesn't make sense to me
Coz, in the case that that would throw, there are no readable bytes left
Meaning that the next method after that would throw
// Waterfall start
if (input.readableBytes() == 0) {
throw new BadPacketException("No more bytes reading varint");
}
// Waterfall end
in = input.readByte();
well i definitely dunno
lean gobletBOT
yeah I'm seeing that too on the latest Waterfall build, gonna have to look at what changed there
void lion
Hello ! Do you know how can i modify the list of commands sent to players when they log in? On paper we can use the PlayerSendCommandEvent but on Watefall i don't found anything
trail plume
ref the javadocs, there was a define commands event
void lion
Oh yes thanks !
finite rock
How to programm with paper? Is it like bukkit? "extends JavaPlugin" doesn't work for me
snow bolt
finite rock
oh my fault
outer cosmos
@finite rock yes fully compatible.
boreal python
how do i make where i can have a queue system like 2b2t
molten jackal
How should I go about getting the default server priority? Not based on the player, since I do modify their reconnect server (not sure if that messes with stuff).
I would also prefer not having to go through the ProxyConfig to get it 😛
trail plume
No
molten jackal
Quite a shame
trail plume
The config is the "live" object
molten jackal
👍
Is there a reason there's no getServerPriority or something in the ReconnectHandler?
It just stores the data for the individual players I see, but the thing is, it has to get the default option somewhere, right?
trail plume
Yea, it gets the default from the listenerinfo config
Oh, actually, false
It doesn't get the default, it just returns null if there is no specific server it should send them to
molten jackal
Ok. So if I use ProxiedPlayer#setReconnectServer(String) and supply null as the argument, it would reset them to the default server?
trail plume
Yes
data.put( key( player ), ( player.getReconnectServer() != null ) ? player.getReconnectServer().getName() : player.getServer().getInfo().getName() );
It's scary how gross bukkit is
I mean, bungee
molten jackal
Ok cool. Is it possible to supply null over a plugin message as a UTF and the bungee still reads that input as null?
molten jackal
Yeah both 😛
trail plume
Actually, wait, setting that to null won't do anything for you
thet check is literally a not null
jesus my brain is zapped
trail plume
That's just a gross way to override the server that's saved by the reconnecthandler
molten jackal
well never mind the get default server method in the player's listener is depreciated
trail plume
Basically, if you wanna override that behavior, replace the reconnect handler yourself
molten jackal
sender.getPendingConnection().getListener().getServerPriority()
Uhm
Ya?
I don't know if this is overridden by the setReconnectServer method
being that setting the reconnect server puts that new server at the top of that priority list
trail plume
It doesn't
molten jackal
so that getServerPriority() method should always work from the default in the config then?
trail plume
On connection, the proxy checks if the reconnection handler has a server that they should connect to
molten jackal
(given i dont modify it)
trail plume
Otherwise, it goes to the logic of using the priorities
and yea, that list is literally backed from the config
molten jackal
I see
so... if I'm gathering this all correctly...
sender.setReconnectServer(getInstance().getProxy().getServerInfo(sender.getPendingConnection().getListener().getServerPriority().get(0)));
should set the reconnect server of the player to the default server in the list? or should i just set their reconnect server in this case to null so it defaults to the priority list instead
trail plume
setting it to the 0th server would break the priorities iirc
molten jackal
Right
What about null though? That should make it use the priorities instead, right?
trail plume
No
Because you can't set that to null, basically
null means "don't override the server that we save"
molten jackal
So... then I can't reset someone's reconnect server after setting it?
trail plume
The only way to get true defined behavior from the reconnection stuff is to define your own
molten jackal
Hm
trail plume
That handler is basically one of the many major fuckups and "special" designs of bungee
molten jackal
Yaaaaahhhh
Is there a default ReconnectHandler I can extend so that I don't have to do all the close() and save() stuff?
like just super it
trail plume
Because it's in a module, not in the API
molten jackal
I see.
so why do they use raw types and not even suppress the warnings?
.<
looking at the source code for ProxiedPlayer, it looks like ProxiedPlayer#getReconnectServer() does return null if it is defaulting to the priority list instead.
that leads me to believe setReconnectServer(ServerInfo) can accept a null argument...
trail plume
That literally just sets a field on ProxiedPlayer
if that field is null, the handler just saves their current server
molten jackal
side question (but related), if force_default_server in the config is true, does that mean the reconnect server field will not do anything?
trail plume
correct