TickThreading - concurrent entity/tile entity ticks and other optimisations
- Thread starter nallar
- Start date
Looking for a bit of advice/direction. I am having issue with my server (excess memory use/growth) and I have come here as almost a last resort. I can't find anything in MCEdit, I am new to it, so I may be missing something. I ran the aforementioned mod and have some results that I am unsure on how to properly asses. Any info is greatly appreciated.
http://pastebin.com/W2jck4v0
http://pastebin.com/W2jck4v0
Add the "-XX:+HeapDumpOnOutOfMemoryError" and then a "java_<something>.hprof" file will be created in your server directory when it runs out of memory. You can then upload that file somewhere and I'll look at it with jvisualvm to see what's using all the RAM when it crashes.
edit: Also, it's 5AM and I need to sleep. If anyone else here knows how and can be bothered, please help goOutSide.
I have manually added "-XX:+HeapDumpOnOutOfMemoryError" to a shell call; unfortunately, this avoids the controls and the CreeperPanel at creeper host. Spent an hour or so looking for the pre-defined startup and couldn't find it. Once this pops I will report the out of memory error.
My server keeps crashing with this error: http://pastebin.com/EiAuA6L6
I really like the plugin, it works excellent, but it's unusable when the server crashes every 10-20 minutes. :L
I'm running the latest build, but this happened on the recommended build as well which is why I upgraded.
I really like the plugin, it works excellent, but it's unusable when the server crashes every 10-20 minutes. :L
I'm running the latest build, but this happened on the recommended build as well which is why I upgraded.
Posted this to the bug tracker already.
Getting this error spammed in my console. Makes any items on the ground unable to pick up, thus we restart. Temporarily fixed by setting world ticking to false.
http://pastebin.com/hi64XFjh
EDIT: running version 732 with forge version 515.
Getting this error spammed in my console. Makes any items on the ground unable to pick up, thus we restart. Temporarily fixed by setting world ticking to false.
http://pastebin.com/hi64XFjh
EDIT: running version 732 with forge version 515.
I tried setting world ticking to false and for some reason it still crashes with the same error.
All I can do is uninstall and wait for a fix I suppose...
The latest build is currently rather experimental. It may fix the exceptions you had there pherce, but not the crash - in fact it looks like the server didn't crash, just IC2 took far too long for a network update for some reason. :s
I'm not sure how the TickNextTickList is getting out of synch, I'll try to fix it soon.
I'm not sure how the TickNextTickList is getting out of synch, I'll try to fix it soon.
Try restarting, it will probably fix it. Sorry that that's required, still haven't discovered why that happens.
tried a restart but enet goes down as soon as i start the server
. No eu flowing at allEncountered a pretty bad bug on my server with TickThreading.
No stack traces, but every time the server goes down every redstone energy cell (Thermal Expansion) gets set to 0 MJ stored.
It's a real shame, we were really enjoying the awesome performance improvements, but that bug is too much =\
I've also noticed that, should a tick event have an exception when the server starts, the memory usage of my server very quickly spirals out of control and has to be shut down. We're talking from ~750MB to 4.5GB in just a few minutes, with no server activity at all. Neither of the above effects were encountered without TT installed.
No stack traces, but every time the server goes down every redstone energy cell (Thermal Expansion) gets set to 0 MJ stored.
It's a real shame, we were really enjoying the awesome performance improvements, but that bug is too much =\
I've also noticed that, should a tick event have an exception when the server starts, the memory usage of my server very quickly spirals out of control and has to be shut down. We're talking from ~750MB to 4.5GB in just a few minutes, with no server activity at all. Neither of the above effects were encountered without TT installed.
Can you make an issue with a heap dump from when it runs out of memory?
Which build were you testing when redstone energy cells broke?
I've added "-XX:+HeapDumpOnOutOfMemoryError" to the recommended parameters. If you're not already using it, please do, it makes the jvm write a heap dump if it runs out of memory. That can the be given to me, and I can find what's using all your RAM.
edit 2: Maybe that's not so useful, the heapdump will end up being huge.
edit: I've added the option to clear world objects on unload. This should mostly fix some memory leaks caused by mods holding on to world objects.
edit 2: Maybe that's not so useful, the heapdump will end up being huge.
edit: I've added the option to clear world objects on unload. This should mostly fix some memory leaks caused by mods holding on to world objects.
I've made some changes in latest which should fix enet. You may still get some error spam - that's normal on the version of IC2 used in the mindcrack pack - but it should at least let current flow.