Solved Mindcrack 8.2.0 Server increasing lag - HELP please

thr33p10

New Member
Jul 29, 2019
26
0
0
Since Mindcrack 8.x there have been some problems with lag increasing with uptime, but nothing like what I've seen with the latest version, 8.2.0. Unfortunately the problems get so bad as to be mostly unplayable due to block lag, in less than 6 hours of uptime.

As an example, the FML log from startup to shutdown where, by the time it shut down over 1.3 GB of RAM had been allocated, for no good reason I could come up with, and there are constant (every minute) "can't keep up..." messages.


2013-03-14 05:39:46 [FINE] [XyCraft] Unloading MultiBlock: 376, server
2013-03-14 05:39:46 [FINE] [XyCraft] Deferring MultiBlock: 377, server
2013-03-14 05:39:48 [FINE] [XyCraft] Adding MultiBlock: 377, server
2013-03-14 05:40:03 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:40:22 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:40:46 [FINE] [XyCraft] Unloading MultiBlock: 377, server
2013-03-14 05:40:46 [FINE] [XyCraft] Deferring MultiBlock: 378, server
2013-03-14 05:40:46 [FINE] [XyCraft] Adding MultiBlock: 378, server
2013-03-14 05:40:54 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:41:17 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:41:47 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:41:49 [FINE] [XyCraft] Unloading MultiBlock: 378, server
2013-03-14 05:41:49 [FINE] [XyCraft] Deferring MultiBlock: 379, server
2013-03-14 05:41:49 [FINE] [XyCraft] Adding MultiBlock: 379, server
2013-03-14 05:42:16 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:42:48 [FINE] [XyCraft] Unloading MultiBlock: 379, server
2013-03-14 05:42:48 [FINE] [XyCraft] Deferring MultiBlock: 380, server
2013-03-14 05:42:48 [FINE] [XyCraft] Adding MultiBlock: 380, server
2013-03-14 05:43:48 [FINE] [XyCraft] Unloading MultiBlock: 380, server
2013-03-14 05:43:48 [FINE] [XyCraft] Deferring MultiBlock: 381, server
2013-03-14 05:43:48 [FINE] [XyCraft] Adding MultiBlock: 381, server
2013-03-14 05:43:56 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:44:39 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:44:49 [FINE] [XyCraft] Unloading MultiBlock: 381, server
2013-03-14 05:44:49 [FINE] [XyCraft] Deferring MultiBlock: 382, server
2013-03-14 05:44:49 [FINE] [XyCraft] Adding MultiBlock: 382, server
2013-03-14 05:45:01 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:45:18 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
2013-03-14 05:45:33 [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?

So .. is the update to xycraft a potential cause (no players were online during this snippet of time in the log)?

No crashes fortunately, but unplayable is still bad for a server...
Just a little extra info: we are running the Mindcrack 8.2.0 pack, and the server has added mods PlayerAPI 1.8, Smartmoving 13.1, and Mystcraft 0.10.1

Thanks to any pointers or help.
 

BuzZBladE

New Member
Jul 29, 2019
35
0
0
Literally billions of commands are flying in the background whilst running all the operations of the server. The console just shows important stuff. One line is next to another that does not mean they are related. The XyCraft lines are debugging information indicated by [FINE]. These are fine and not errors. It is displayed only as information the developer wants to see to check his work.

Line : [WARNING] [Minecraft] Can't keep up! Did the system time change, or is the server overloaded?
Because the server is overloaded. This is telling you your TPS is dropping way below 20 tick cycles per second.

If you are out of memory, it will just shut down completely and say, Hey, I am out of memory that was allocated for me. Memory size is not the issue but rather the rate at which it can process. CPU.

That being said. There are little things like adding -nogui to the java's startup batch command line when starting the server.

Do you have access to the startup command for the server?
 

thr33p10

New Member
Jul 29, 2019
26
0
0
Between posting this and reading a slightly older thread.. I think the problem may be solved. See this post: http://forum.feed-the-beast.com/threads/mindcrack-server-lag.7704/ for the suggestions I tried out on changing the .bat file for launching the server jar. It dropped my CPU load by 50% or more!!!

Here's my video link showing a playtest of trying to get the server to complain at me:

Anyway, tell me if I'm crazy for just trying out the suggested options for Java.. I'll know after the next 24 hrs of uptime if this is an improvement or just the start of a different "problem"
 

thr33p10

New Member
Jul 29, 2019
26
0
0
Alright, there are still some problems with memory leaks somewhere...

Despite the fantastic performance increase that I have seen with the java options I have used, it eventually runs the program to be using all allocated RAM now.. and then runs into the "out of heap" error.
EDIT: pastebin of crash report: http://pastebin.com/cj6zGSwv

Is there anyone who could help with fine-tuning the Java flags that are used to call the server?

The old setup I used which gave the previous problems of staying alive long, but unplayable:

java -server -Xms512M -Xmx3G -XX:+AggressiveOpts -jar ftbserver.jar


The experimental options that made it look like minecraft was amazingly optimized for a server:

java -server -Xincgc -Xmx3G -Xms1G -Xmn512M -XX:NewRatio=2 -XX:CMSFullGCsBeforeCompaction=1 -XX:SoftRefLRUPolicyMSPerMB=2048 -XX:+CMSParallelRemarkEnabled -XX:+UseParNewGC -XX:+DisableExplicitGC -Xnoclassgc -XX:+UseFastAccessorMethods -XX:CMSInitiatingOccupancyFraction=90 -XX:+UseConcMarkSweepGC -XX:UseSSE=3 -XX:+UseCMSCompactAtFullCollection -XX:parallelGCThreads=2 -XX:+AggressiveOpts -XX:ReservedCodeCacheSize=1G -cp ftbserver.jar -jar ftbserver.jar

I have increased the flax "-Xmx3G" to "-Xmx4G" for the moment, but I wonder if there's any other tweaking that would help...
 

thr33p10

New Member
Jul 29, 2019
26
0
0
Problem more concisely identified... Java start options are about as optimized as they will ever be. The problem is somewhere on the server map :( :( :(

Time to hunt down how to figure out where on the map the lag is coming from..