Problems with DH pregen: RAM usage, overloaded DH que, overloaded server
log: https://mclo.gs/9szT28I
System RAM: 32 GB ddr5
Server allocated RAM: 25 GB
CPU: i7-13700F
I left DH pregen to run overnight, did not go well...
some minutes after generation started, I could observe warning messages such as this one:
[21:51:03] [Server thread/WARN]: §6Distant Horizons overloaded, too many chunks queued for LOD processing. §r
This may result in holes in your LODs.
Fix: move through the world slower, decrease your vanilla render distance, slow down your world pre-generator (IE Chunky), or increase the Distant Horizons' CPU thread counts.
Max queue count [15000] ([1000] per thread+players).
They started at 21:51 and mysteriously stopped at 05:31, around when cps dropped below 70. These warnings occur regardless of how many threads I have allocated to DH, even when I allocated 22 of my 24 threads
At 6:58 I could observe warning messages such as this one:
[06:58:57] [Server thread/WARN]: Can't keep up! Is the server overloaded? Running 2472ms or 49 ticks behind
They occured untill I closed the server
During the whole process ram usage steadily increased, cps going down as it did
I can observe this phenomenon in every pregen I run in this server.
14 Replies
when I stopped the server, after having stopped pregen, There were warning messages such as this one:
[11:26:32] [DH-PriorityTaskPicker Thread[9]/WARN]: DistantHorizons: Couldn't load chunk [[1467, -685]] from server, error: [Unloaded chunk].
java.lang.RuntimeException: Unloaded chunk
at knot/loaderCommon.fabric.com.seibel.distanthorizons.common.wrappers.worldGeneration.BatchGenerationEnvironment.lambda$requestChunkFromServerAsync$23(BatchGenerationEnvironment.java:905) ~[DistantHorizons-2.3.4-b-1.21.5-fabric-neoforge.jar:?]
at knot/net.minecraft.class_9259$class_9260.method_57132(class_9259.java:104) ~[server-intermediary.jar:?]
at knot/loaderCommon.fabric.com.seibel.distanthorizons.common.wrappers.worldGeneration.BatchGenerationEnvironment.lambda$requestChunkFromServerAsync$24(BatchGenerationEnvironment.java:905) ~[DistantHorizons-2.3.4-b-1.21.5-fabric-neoforge.jar:?]
at java.base/java.util.concurrent.CompletableFuture$UniApply.tryFire(CompletableFuture.java:646) ~[?:?]
at java.base/java.util.concurrent.CompletableFuture.postComplete(CompletableFuture.java:510) ~[?:?]
at java.base/java.util.concurrent.CompletableFuture.completeExceptionally(CompletableFuture.java:2194) ~[?:?]
at knot/com.ishland.flowsched.scheduler.ItemHolder.setStatus(ItemHolder.java:214) ~[c2me-base-0....+1.21.5-eb5348114a2b4691.jar:?]
at knot/com.ishland.flowsched.scheduler.StatusAdvancingScheduler.tick(StatusAdvancingScheduler.java:180) ~[c2me-base-0....+1.21.5-eb5348114a2b4691.jar:?]
at knot/com.ishland.c2me.rewrites.chunksystem.common.TheChunkSystem.tick(TheChunkSystem.java:222) ~[c2me-rewrites-chunk-system-0....+1.21.5-c4e34796a48e34d1.jar:?]
at knot/com.ishland.flowsched.scheduler.DaemonizedStatusAdvancingScheduler.pollTasks(DaemonizedStatusAdvancingScheduler.java:82) ~[c2me-base-0....+1.21.5-eb5348114a2b4691.jar:?]
at knot/com.ishland.flowsched.scheduler.DaemonizedStatusAdvancingScheduler.mainLoop(DaemonizedStatusAdvancingScheduler.java:43) ~[c2me-base-0....+1.21.5-eb5348114a2b4691.jar:?]
at java.base/java.lang.Thread.run(Thread.java:1583) [?:?]
...this one:
[11:26:38] [Server thread/WARN]: World generator thread pool shutdown didn't complete after [3] seconds. Some world generator requests may still be running.
...and 4 of this one:
[11:26:39] [Server thread/WARN]: Failed to generate section 0*0,0
Does anyone know what is happening?
Maybe it is a mod compatibility issue?
Maybe I did something wrong?
Help is appreciated...
What did you use for the pregen?
what do you mean?
the command was dh pregen start minecraft:the_nether 0 0 1984
Was anyone online during the pregen?
I don’t think queue error should pop up during pregen though I heve never myself used it
What generation mode did you use?
Do you have any worldgen or structure mods or datapacks?
You can speed the pregen up and increase the queue size by increasing the DH CPU load preset
You can reduce the server load possible fixing the tps and tick issues by decreasing the DH CPU load preset
I have amplified nether
isn't cpu load preset just presets for how many threads you give DH, like the dedicated option?
also can I just ignore the holes in LODS warning and just run the pregen again to fill up these holes?
no
full/internal_server
oh so too low means que issues and too high means general performance issues... makes sense...
but it seems 22 threads are not enough to fix que issues💀
The dedicated thread + run time ration of those threads, idk if any other settings get changed too
That should work
Do you have C2ME installed?
yea
also are these holes in LODS or vanilla chunks?
With FULL gen mode?
I would guess both, though it is possible they are LoD only
ok...
do you know anything about the RAM issue or the errors?
(btw ping me so I answer)
No idea about the RAM climbing up
also seems weird that it had to unload cchunks...
other than Herobrine being logged on flying around loading chunks and filling up DH que I can't think of anything...
BTW what is thread run time ratio?
A ratio between the thread running and sitting dormant
E.G. if the ratio is at 0.6 then if the last task took 30 s it should wait 20 s before starting the next task
Or at least that is the gist of it, no idea how it works exactly, does it pause the task, etc.
Hmm what ratio would you recommend for this?
1.0