Issue type:

• 🐌 Performance issue

Short description:

Connecting 4 networks with 5 exporters each lead to an extreme lag, when first discovered, it required us to load a backup that was made before connecting them, because loading 1 chunk took like 20 seconds
Also it's in the ATM 10 2.34 modpack

Steps to reproduce the problem:

1. Repeat setup in the screenshot, each exporter has a list with 4 items, blacklist enabled, passive interaction doesn't matter, tested with both enabled and disabled
2. then connect those 4 networks

Versions:

• This mod: 1.25.3
• Minecraft: 1.21.1
• Forge: NeoForge 21.1.116

Profiler output:

I can't upload sparkprofile, so here's a screenshot of relevant info

Also it's 100% reproducable, I can also reproduce it from 0 after restoring backup, and profiler results looks almost the same

Thanks for reporting!

well even after separating it's lagging, but not as much

exporters are causing the lag

This is similar to reports #1460 and #1470.

What are you exporting into exactly? Does the lag also occur when exporting into some other inventory?

What do the lists look like that you are using to export? Are they static or dynamic?

What are you exporting into exactly? Does the lag also occur when exporting into some other inventory?

What do the lists look like that you are using to export? Are they static or dynamic?

into loot fabricators, those orange blocks, but doesn't matter really, any inventory will work, also every connected block was empty (simulation chambers too)
List was created using programmer, I just opened programmer, selected "List", and started adding items there, in total 4 items, they are all general predictions: twilight prediction, overworld prediction, nether prediction, end prediction.

into loot fabricators, those orange blocks, but doesn't matter really, any inventory will work, also every connected block was empty (simulation chambers too)

That's odd. I've not seen any issues recently with normal inventories.

Could you try uploading the spark logs again? Because I will need some more details that the current screenshot does unfortunately not provide.

pxtUecYv1h.txt
rename to something.sparkprofile

into loot fabricators, those orange blocks, but doesn't matter really, any inventory will work, also every connected block was empty (simulation chambers too)

That's odd. I've not seen any issues recently with normal inventories.

Could you try uploading the spark logs again? Because I will need some more details that the current screenshot does unfortunately not provide.

also, what's interesting, is that /neoforge tps will report nice tps, but in spark it would show really bad tps, the real one I think, so neoforge doesn't see that for some reason

Ok, this is weird. For some reason our cache (which is supposed to optimize things) is causing lag for you.
More specifically, the system call vdso.__vdso_clock_gettime() is extremely slow:

I've never seen this before tbh.

Any idea what this could be about? Are you running your server on some exotic OS by any chance?

I also was looking at profiler, noticed this, and was like "huh?"

Ok, this is weird. For some reason our cache (which is supposed to optimize things) is causing lag for you. More specifically, the system call vdso.__vdso_clock_gettime() is extremely slow:

I've never seen this before tbh.

Any idea what this could be about? Are you running your server on some exotic OS by any chance?

server is running on NixOS 24.11.20250126.4e96537 (Vicuna) x86_64, Java 21 from OpenJDK, Linux 6.6.72

server is running on NixOS 24.11.20250126.4e96537 (Vicuna) x86_64, Java 21 from OpenJDK, Linux 6.6.72

I guess that should be fine 🤷

In any case, I don't think this is a mod issue, but a system/OS issue.
But I'm afraid I can't help you any further on solving it :/

I also was looking at profiler, noticed this, and was like "huh?"

That could also explain the TPS inconsistencies. As the lag here is not coming from Java, but from native calls.

This could be a related issue: raspberrypi/linux#6601

very very strange

yeah, can't replicate in singleplayer, it's not a mod issue. I will try some fixes and will comment them there, but I'm closing the issue, no reason to keep it open.

@rubensworks fixed by updating to latest linux kernel instead of LTS kernel

Aha, nice to hear the problem has been fixed!

@rubensworks we migrated to other host, and problem appeared again, but now it's not in native libraries, but in com.google.common.collect.TransformedIterator.hasNext(), or net.minecraft.world.item.ItemStack.().
and we're not the only one who got this, I also checked ATM10 discord server for this, and it appears there are also other people with such problems, here's a profile of one of them for example: https://spark.lucko.me/OnrPfGsloh, this is the message link if you're interested: https://discord.com/channels/254530689225981953/1255043369625976863/1326645651555090463
Something is really wrong in ID, as it doesn't happen with other pipe mods AT ALL, and enabling safe mode in ID ofc fixes this completely
I really like ID, and I want to play with it, but this is just bad

spark

spark is a performance profiler for Minecraft clients, servers, and proxies.

Discord

Discord - Group Chat That’s All Fun & Games

Discord is great for playing games and chilling with friends, or even building a worldwide community. Customize your own space to talk, play, and hang out.

and our profile https://spark.lucko.me/Es89equ1l5, although the critical, laggiest point/function/call changes sometime

spark

spark is a performance profiler for Minecraft clients, servers, and proxies.

The first log is caused by Sophisticated Storage (duplicate of #1460), which is followed up in P3pp3rF1y/SophisticatedStorage#584.
The second log seems to be caused by Functional Storage, with probably the same root cause as #1460. Could you report this to their issue tracker and link back to here and #1460?