gc issues
hey, i recently updated to 1.21.10 and notice a very high gc time and entity ticking, i cant find the cause of it. ive given the server 16gb ram total
ticks over 120 https://spark.lucko.me/RdF1hCvcGf
all ticks https://spark.lucko.me/hBTULjTPYb
my startup script: https://mclo.gs/Hv2lbnU
21 Replies
Could try using ZGC
i tried it before but it was not working at all, could try it again tho
edit:
yeah zgc makes it way worse
here is another spark report https://spark.lucko.me/8SV9YaIOLD
spark
spark is a performance profiler for Minecraft clients, servers, and proxies.
is this in a VM, docker container or both?
pterodactyl
so docker ig
do you have access to the host machine?
yes
and you're 100% sure it's running bare metal and not in a virtual machine?
hostnamectl status should have a Virtualization line if it is a vmIcon name: computer-vm
Chassis: vm 🖴
make sure your hypervisor isn't running out of memory and using swap
if you can
or check for cpu steal in top
how would i do that? im not too advanced is ssh
steal is the st line at the top
0.8, 1.1, 2.3 27.3 (mc lag), 4.8, 14.9, 8.4, 0.2 (mc lag over), 13
see if it goes up to a high number when your gc takes forever
if it does, complain to whoever sold you the vm
it's out of 100
hm, doesent seem to be the issue, very big lagspike but 0.1 steal, tho the steal was never so consistent on 0.1 when i dont have lag spikes
spark
spark is a performance profiler for Minecraft clients, servers, and proxies.
now thats definitely screaming gc issues https://spark.lucko.me/6o54xasyLd
spark
spark is a performance profiler for Minecraft clients, servers, and proxies.
you can do
/spark profiler start --alloc to start the profiler and profile memory allocations (memory pressure) instead of CPU usage
I will note, someone has an army of 130 villagers and that likely doesn't helpSolution
ye, i believe its just hoster problems causing this
this one didn't even have gc issues (allegedly), just 220 villagers
