Mysterious TPS issues tied to a specific region
Recently we've started having massive server performance issues when a specific player's base is loaded. The ram usage does not go super high, but the TPS becomes terrible and I think the server starts running GC every 30 seconds instead of every 5 minutes. He doesn't have a ton of create machines or anything that would obviously be straining the server at his base. He has a large number of Fossils and Archeology creatures, but not so many that one would assume it would be causing the server to completely freak out whenever the area is loaded.
Attached are a couple spark profiles I've taken recently- does anyone know what might be causing the issue? I thought at first it was the F&A animal AI, but we tried unloading his base, then manually spawning a bunch of F&A creatures and when we did that it didn't seem to lag much. So I didn't think it was that. Any help is greatly appreciated!
https://spark.lucko.me/UaTJmOXJIP
https://spark.lucko.me/tNM4GEWsHl
https://spark.lucko.me/ht0NgGQswa
https://spark.lucko.me/feTMU0ny0x
6 Replies
I don't believe this issue is memory related. Even if it was, a 10ms pause from GC every 30 seconds wouldn't make a noticeable impact in average TPS.
Do you have a normal spark profile instead of an alloc profile?
I don't, what will that tell me instead?
cpu usage, as memory usage shouldn’t be problematic
ok, I'll take one later today and let you know. I should have mentioned I figured it probably wasn't memory related since the memory usage has not been very high
https://spark.lucko.me/ane7q9Kl3p
aside from an above average amount of redstone, nothing jumps out.
have you checked your host’s status for that node and made sure that the node itself isn’t under high cpu usage (or something like that)?
it looks like it's fine(?)
at least from what I can tell
@Snow Kit took a new one https://spark.lucko.me/8jAf84jvUQ