Modpack forge server gives out errors (1.18.2)

[Spezificartz]

I'm playing the modpack better Minecraft plus with some friend on a server with 6g ram and forge version 40.2.0. The server starts successful but gives out many errors at startup. Sometimes when changing dimensions or when 2 players are joining at once for example the server timeouts the client or it just freezes completely. I've tried installing some performance mods and preloaded the chunks with "chunky" it helped a bit but I'm still getting timeouts sometimes. I've uploaded the server log to https://mclo.gs/sik1eFz

Would be nice if someone could help me a 

 

 

 

[warjort]

Use a mod like this to analyze what is really happenning.

https://www.curseforge.com/minecraft/mc-mods/spark

[Spezificartz]

8 hours ago, warjort said:

Use a mod like this to analyze what is really happenning.

https://www.curseforge.com/minecraft/mc-mods/spark

I already have spark installed and ran some monitoring tasks but couldn´t find a fix. 

Log: https://spark.lucko.me/ekqUw2RAa9?hl=25,26,27

[warjort]

I am no expert on reading the spark reports. You should use the spark discord.

For what it's worth.

I don't see anything related to what you describe. i.e. players logging in, or the server thread getting stuck for a long period time (freeze).

If that profile is really representative of what is happening the problem must be elsewhere, e.g. at the networking or system level.

 

According to the metrics, you ran that profile for 1 minute and 1 seconds.

During that time it recorded a full system garbage collection taking 1600ms (1.6 seconds) and says your memory (swap) is 93% full.

Both are highlighted in red.

Like I said above, the spark developers would be the people to ask on how to interpret their data.

[Spezificartz]

4 hours ago, warjort said:

I am no expert on reading the spark reports. You should use the spark discord.

For what it's worth.

I don't see anything related to what you describe. i.e. players logging in, or the server thread getting stuck for a long period time (freeze).

If that profile is really representative of what is happening the problem must be elsewhere, e.g. at the networking or system level.

 

According to the metrics, you ran that profile for 1 minute and 1 seconds.

During that time it recorded a full system garbage collection taking 1600ms (1.6 seconds) and says your memory (swap) is 93% full.

Both are highlighted in red.

Like I said above, the spark developers would be the people to ask on how to interpret their data.

The problem is that when im currently running the profiler and the server freezes/crashes, i´m not able to save/upload the  profiler log bc the server just stops. But anyways, thanks for the reply!

[warjort]

You can try to capture a thread dump when it freezes to see if it identifies where the server is getting "stuck".

You will need a copy of the java development kit to do it.

https://www.baeldung.com/java-thread-dump#:~:text=jstack is a command-line,its output to a file.

[Spezificartz]

4 hours ago, warjort said:

You can try to capture a thread dump when it freezes to see if it identifies where the server is getting "stuck".

You will need a copy of the java development kit to do it.

https://www.baeldung.com/java-thread-dump#:~:text=jstack is a command-line,its output to a file.

Okay thx I'll try that. I tested the same modpack(some mods are different) on version 1.16.5 and it runs perfectly fine with a much better tps, maybe I just change to 1.16. I've heard that some people got performance problems especially after updating to 1.18 

Edited January 20, 2023 by Spezificartz