Resolve memory usage issues related to Netty 4.2 update

[onebeastchris]

Should fix #5467, fix #5468 (the actual bungee/velocity issues reported on there; the issue on geyser-spigot by the original author seems completely unrelated), fix #5482, fix #5510

These builds will also require using the latest BungeeCord & Velocity builds to properly work!

You can download the new builds here:

• Geyser-Bungeecord.jar
• Geyser-Fabric.jar
• Geyser-NeoForge.jar
• Geyser-Spigot.jar
• Geyser-Standalone.jar
• Geyser-Velocity.jar
• Geyser-ViaProxy.jar

TODO:

• deal with io_uring not being experimental anymore

[MatinHQ]

Any ETA when it's get merge into main branch?

[onebeastchris]

As soon as enough people report that it's working! So far, we've had no responses.

[MatinHQ]

As soon as enough people report that it's working! So far, we've had no responses.

[00:52:24 ERROR] [/188.214.172.18:18460] Error while ticking RakSessionCodec state=CONNECTED channelActive=true
java.lang.OutOfMemoryError: Cannot reserve 131072 bytes of direct buffer memory (allocated: 4294886160, limit: 4294967296)
        at java.nio.Bits.reserveMemory(Bits.java:178) ~[?:?]
        at java.nio.DirectByteBuffer.<init>(DirectByteBuffer.java:111) ~[?:?]
        at java.nio.ByteBuffer.allocateDirect(ByteBuffer.java:360) ~[?:?]
        at io.netty.buffer.UnpooledDirectByteBuf.allocateDirect(UnpooledDirectByteBuf.java:104) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.UnpooledDirectByteBuf.<init>(UnpooledDirectByteBuf.java:64) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.UnpooledUnsafeDirectByteBuf.<init>(UnpooledUnsafeDirectByteBuf.java:41) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.UnsafeByteBufUtil.newUnsafeDirectByteBuf(UnsafeByteBufUtil.java:687) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptiveByteBufAllocator$DirectChunkAllocator.allocate(AdaptiveByteBufAllocator.java:118) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptivePoolingAllocator$Magazine.newChunkAllocation(AdaptivePoolingAllocator.java:763) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptivePoolingAllocator$Magazine.allocate(AdaptivePoolingAllocator.java:688) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptivePoolingAllocator$Magazine.tryAllocate(AdaptivePoolingAllocator.java:567) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptivePoolingAllocator.allocate(AdaptivePoolingAllocator.java:234) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptivePoolingAllocator.allocate(AdaptivePoolingAllocator.java:220) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AdaptiveByteBufAllocator.newDirectBuffer(AdaptiveByteBufAllocator.java:70) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AbstractByteBufAllocator.directBuffer(AbstractByteBufAllocator.java:188) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AbstractByteBufAllocator.directBuffer(AbstractByteBufAllocator.java:179) ~[Geyser-Standalone.jar:?]
        at io.netty.buffer.AbstractByteBufAllocator.ioBuffer(AbstractByteBufAllocator.java:140) ~[Geyser-Standalone.jar:?]
        at org.cloudburstmc.netty.handler.codec.raknet.common.RakSessionCodec.onTick(RakSessionCodec.java:450) ~[Geyser-Standalone.jar:?]
        at org.cloudburstmc.netty.handler.codec.raknet.common.RakSessionCodec.tryTick(RakSessionCodec.java:405) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.PromiseTask.runTask(PromiseTask.java:98) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.ScheduledFutureTask.run(ScheduledFutureTask.java:162) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.AbstractEventExecutor.runTask(AbstractEventExecutor.java:148) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:141) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:507) ~[Geyser-Standalone.jar:?]
        at io.netty.channel.SingleThreadIoEventLoop.run(SingleThreadIoEventLoop.java:180) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.SingleThreadEventExecutor$5.run(SingleThreadEventExecutor.java:1073) ~[Geyser-Standalone.jar:?]
        at io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74) ~[Geyser-Standalone.jar:?]
        at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30) ~[Geyser-Standalone.jar:?]
        at java.lang.Thread.run(Thread.java:1583) [?:?]
[00:52:25 ERROR] [/5.125.216.210:55858] Error while ticking RakSessionCodec state=CONNECTED channelActive=true

Tested that on geyser-standalone and it almost allocated 9GB of ram in 5 minutes with 30 players connected and XMX was 4GB

[onebeastchris]

Can you provide a geyser dump?

[onebeastchris]

@MatinHQ should be resolved, thanks for the report.
For other platforms:

• Velocity is setting the pooled allocator globally since this commit: PaperMC/Velocity@dc65953; fixing the issue
• I've created a PR for similar behavior on BungeeCord: Set allocator using netty system property to apply pooled allocator globally SpigotMC/BungeeCord#3828 - if this PR isn't accepted we'll need to make further changes in our Bedrock protocol library dependencies to fully resolve the issue.

[MatinHQ]

I tested latest commit it's much better but problem still exist I messaged you on discord for geyser dump for privacy reasons I can't share it here

[astirling]

This build does not have the LocalServerChannel error, or any other error on load, that I reported

BungeeCord: b1948