[Crash]: Severe TPS spikes after updating ATM 10 from 4.6 → 4.9 (roll-back fixes it)
Closed this issue · 10 comments
Possible Fixes
Yes
Modpack Version
4.9
Crash Assistant Message (big green button)
No response
What happened?
After updating our server from 4.6 to 4.9, we started getting random, long freezes (30+ seconds). Console “can’t keep up” warnings jumped from ~2000ms (expected/understood on our setup) to 10000–50000ms. Rolling back to 4.6 immediately restores normal behavior with the usual steady ~2000ms warnings and no long freezes.
Environment
Modpack: ATM 10
Versions tested: 4.6 (stable), 4.9 (problem), 4.8 (problem, logs attached)
OS/Host: [AMD Ryzen 5 5500U/16GB/SSD M.2 7000/7000]
Network on Host PC: 400 mb/s | 340mb/s
Player count during tests: 2
View distance / simulation distance: 12 chunks
What changed:
Only the modpack version (4.6 → 4.9).
Configs, world, and forced chunk setup remained the same.
Symptoms on 4.9
Random freezes lasting 30+ seconds.
Console warnings escalate dramatically:
Can't keep up! Is the server overloaded? Running 10000ms–50000ms behind
Occasional memory leaks observed on 4.9. We do not see memory leaks on 4.6.
Refined Storage controller buffer/throughput behavior changed after the update:
On 4.6: controller sits steady at ~50k FE.
On 4.9: controller fluctuates around 17k–25k FE, even though our autocrafting/energy setup is unchanged.
Logs sometimes blame “too many chunks loading”, but:
We have the same forced chunks on 4.6 and it’s fine there.
Notes & things we tried
We know about one farm that causes the steady ~2000ms “can’t keep up” on 4.6; it never causes noticeable lag spikes for players there.
On 4.9 we saw log lines about Endermen trying to place/break blocks in a claimed area. We found and killed them; issue persisted.
Rolling back to 4.6 immediately removes the freezes; warnings return to the stable ~2000ms baseline.
Why we think this may be a 4.9 regression (or config/mod interaction)
Identical world, forced chunks, and farm behave normally on 4.6.
4.9 introduces:
Much higher “can’t keep up” deltas (10–50 seconds).
Changed RS energy buffer behavior.
Intermittent memory leaks we didn’t see on 4.6.
2025-08-23-7.log
2025-08-23-6.log
crash-2025-08-23_21.52.47-server.txt
2025-08-23-5.log
2025-08-23-4.log
2025-08-23-3.log
2025-08-23-2.log
2025-08-23-1.log
Crash Log or Disconnect Log
Latest.Log
Update (ATM10 4.10): Reproducible post-login stall → both players timed out
When: 2025-08-24 at 19:27:55
What happened: After updating to 4.10, both players joined the server; ~60 seconds after login the server froze with:
Can't keep up! Is the server overloaded? Running 92827ms (1856 ticks) behind
Both clients then timed out. Rolling back to 4.6 removes this behavior (we only see our usual ~2000ms baseline from a known farm, with no player-visible stutter).
Find logs below:
debug-2.log
debug-1.log
2025-08-24-2.log
2025-08-24-1.log
latest (1).log
debug (1).log
You need to spark profiler the server
/spark profiler --timeout 60
after you get the link also run
/observable run 60
share results
on this one it was very idle, maybe you would have to test when having multiple players online
maybe try /spark profiler --timeout 180 --only-ticks-over 75 so we can grab the spikes
lags every 1-3 mins,
https://spark.lucko.me/CG39b55y3M
/spark profiler --timeout 180 --only-ticks-over 75
lags every 1-3 mins, https://spark.lucko.me/CG39b55y3M
/spark profiler --timeout 180 --only-ticks-over 75
It is something on refined storage, investigate what your players tried to craft that can be causing it
@raoulvdberge Do you have any hint on this one on what they can investigate?
Already tracked in refinedmods/refinedstorage2#987
This is an optimization problem with the autocrafting upgrade in an exporter. Remove these for now if they cause problems.