Admincraft•2y ago

Continuous Purpur Thread Dumps, related to memory?

I keep having these weird thread dumps and OOM crashes on my server which are causing it to lock down entirely, requiring me to have to use the kill button in Pterodactyl. I have configured Aikars flags to limit the RAM usage of the server (I pay for 6 GB, Xmx is set to 5GB), and I've tried doing some profiling to no avail. All I know is that each thread dump referenced the UUID of one of our Admins who was just exploring. Any ideas? I'm at a loss, I really don't know what's going on at this point. I thought it may have been BlueMap because it's the most "resource intensive" thing I've done recently but I uninstalled that plugin and still have issues. I've also tried pregenerating chunks with Chunky, but I think that may be making it worse. Plugin List:

Paper Plugins: (2):
[23:30:57 INFO]:  - eco, EconomyShopGUI
[23:30:57 INFO]: Bukkit Plugins: (45):
[23:30:57 INFO]:  - BetterRTP, Chunky, ChunkyBorder, Citizens, CMILib, *CommandSpy, CoreProtect, Crystal-Extra-Flags, DecentHolograms, DimensionPause
[23:30:57 INFO]:  DiscordSRV, DiscordSRV-Staff-Chat, EcoCrates, Essentials, EssentialsChat, EssentialsSpawn, ExtraContexts, FastAsyncWorldEdit, floodgate, Geyser-Spigot
[23:30:57 INFO]:  GriefPrevention, InventoryRollbackPlus, ItemEditor, Jobs, LuckPerms, Maintenance, Multiverse-Core, Multiverse-Inventories, Multiverse-Portals, NBTAPI
[23:30:57 INFO]:  PlaceholderAPI, PlayerWarps, PlugManX, ProtocolLib, QuickShop, sleep-most, spark, TAB, UltimateAutoRestart, Vault
[23:30:57 INFO]:  ViaVersion, Votifier, VotingPlugin, WorldGuard, WorldGuardExtraFlags

Paper Plugins: (2):
[23:30:57 INFO]:  - eco, EconomyShopGUI
[23:30:57 INFO]: Bukkit Plugins: (45):
[23:30:57 INFO]:  - BetterRTP, Chunky, ChunkyBorder, Citizens, CMILib, *CommandSpy, CoreProtect, Crystal-Extra-Flags, DecentHolograms, DimensionPause
[23:30:57 INFO]:  DiscordSRV, DiscordSRV-Staff-Chat, EcoCrates, Essentials, EssentialsChat, EssentialsSpawn, ExtraContexts, FastAsyncWorldEdit, floodgate, Geyser-Spigot
[23:30:57 INFO]:  GriefPrevention, InventoryRollbackPlus, ItemEditor, Jobs, LuckPerms, Maintenance, Multiverse-Core, Multiverse-Inventories, Multiverse-Portals, NBTAPI
[23:30:57 INFO]:  PlaceholderAPI, PlayerWarps, PlugManX, ProtocolLib, QuickShop, sleep-most, spark, TAB, UltimateAutoRestart, Vault
[23:30:57 INFO]:  ViaVersion, Votifier, VotingPlugin, WorldGuard, WorldGuardExtraFlags

Output of /version:

[23:34:09 INFO]: Checking version, please wait...
[23:34:09 INFO]: Current: git-Purpur-2016 (MC: 1.20.1)*
* You are running the latest version

[23:34:09 INFO]: Checking version, please wait...
[23:34:09 INFO]: Current: git-Purpur-2016 (MC: 1.20.1)*
* You are running the latest version

And a snippet of the logs that keep being spit out has been attached as well.

message.txt

40 Replies

Admincraft Meta•2y ago

Please use a paste service instead!

Paste services are more mobile friendly and easier to read than just posting a log file

message.txt

https://mclo.gs/SaXTXcl

Requested by terrorbytetw#0

Admincraft Meta•2y ago

Thanks for asking your question!

Make sure to provide as much helpful information as possible such as logs/what you tried and what your exact issue is

Make sure to mark solved when issue is solved!!!

/close !close !solved !answered

Requested by terrorbytetw#0

TerrorByteOP•2y ago

Thanks buddy ❤️ Currently, my memory usage is stable at 5.01 GB / 6 GB utilized, however when these crashes are happening, memory is spiking to 6.01 and the OOM killer is nuking the server :\ I can provide Spark profiles if needed, as well as full logs and whatever else

shoker ツ•2y ago

Common cases when OOM Killer kills a server: not enough memory for the server OR not enough memory for Java itself. The first case means a memory leak is possible here. Try /spark healthreport when memory usage gets high values to see if it's the server is actually using so much memory, if it is, you'll see a lot of memory being used in healthreport. So if the server really uses so much memory you need to make a /spark heapdump exactly when it uses a lot of memory (almost full), download Eclipse MAT and analyze it there, it has a feature to find memory leaks btw. But it needs manual exploring the dump too. The second case can happen if you allocate a lot of RAM to the server and forget to think about Java itself which needs RAM too. As of my experience it needs like 1GB extra RAM to work fine and not to get killed. Also when using AlwaysPreTouch flag (Aikar's flags have it) the server allocates the whole RAM you gave it at the start making any other process unable to use it if needed, that's where Java process can get hurt. You can try disabling AlwaysPreTouch (just to test it, the purpose of the flag itself is fine) and see if it helps. If it does, it means you need to lower RAM values a bit, maybe 4.5G goes cool, so it needs testing.

Gaming

Programming

Continuous Purpur Thread Dumps, related to memory?

Did you find this page helpful?