Problem Server Lag Issues

[TomPunch]

Forum,

I am attempting to play a FTB Ultimate Server with 1 other person and I am having consider issues with lag (FPS between 2 - 15), even with the lowest video settings. My server is spamming 'Can't Keep Up' warnings, but I have not been able to resolve that issue. I have 8 GB of RAM, and applied 4GB's to FTB.

Any help would be greatly appreciated.

What OS are you running? Windows 7

Are you using minecraft hosting provider or a dedicated/local server not designed for minecraft? Local Server

What version of FTB are you using? (Dont just say latest tell us the version). FTB Ultimate - 1.1.2 with "-XX:MaxPermSize=128m" in the additional options, for Java parameters.

Did you add any mods to the server pack? No

What are your server specs? java -Xms1024M -Xmx4096M -jar ftbserver.jar -nogui

 

[cdauber]

Do you get any error messages like "The world _______ has Leaked" ? I know that was a problem in my server, and I know how to fix it

 

[DZCreeper]

This may not be a server issue, you mentioned low FPS, which is client side.

 

[cdauber]

But the message "can't keep up, did the time system change or is the server overloaded?" Means that the server is having problems

 

[DZCreeper]

Well, can you get some details on the server and client hardware and OS?

If your not sure of those things, this program is easy to use and will show it all to you in a simple to understand fashion.

https://www.piriform.com/speccy

Edit: I assume both the client and server are running Windows, if this is not the case, alternatives do exist for Mac and Linux, but I don't know any at the moment.

 

[TomPunch]

Do you get any error messages like "The world _______ has Leaked" ? I know that was a problem in my server, and I know how to fix it

Now that you have said you have a fix, I wish I did have an issue like that![DOUBLEPOST=1369971989][/DOUBLEPOST]

This may not be a server issue, you mentioned low FPS, which is client side.

I have slowly but surely coming to that conclusion as well. I just can't place my finger on what is causing the issue!

 

[TomPunch]

Well, can you get some details on the server and client hardware and OS?

If your not sure of those things, this program is easy to use and will show it all to you in a simple to understand fashion.

https://www.piriform.com/speccy

Edit: I assume both the client and server are running Windows, if this is not the case, alternatives do exist for Mac and Linux, but I don't know any at the moment.

I am not sure what to tell you about the server, other then it is a standard FTB Ultimate server that I downloaded from the homepage and installed.

The client hardware is windows 7 Enterprise, on a Intel Core 2 Duo CPU 2.80 GHZ with 8 GB's of RAM, and a 64 bit OS.

 

[DZCreeper]

So the server is running on the same machine as the client? Try not to take this hurtfully, but your hardware seems a little underpowered to be running both at the same time. Your also missing details like CPU model number, hard drive model number. Its good to have as many details as possible about hardware so I can pinpoint the exact cause of your problems.

My first suggestion would be replacing the existing ServerStart.bat with this one: https://www.dropbox.com/s/jqrusqvly0hmbho/ServerStart.bat
It may offer a slight increase is server performance. If your still having issues it is likely your client having low FPS. Not much can be done about that aside from new hardware and perhaps trying Optifine. Also, your client should not need more than 1 or perhaps 2 gigs of memory, allocating more could actually decrease performance slightly.

 

[Draylynn]

Working with FTB servers on a daily basis, I'm concerned about the latest updates causing TPS dropping (over time), I've spent hours over the last few days debugging servers trying to find the cause.
Has something updated recently? Auto-updated mod?

On several occasions we've had to reset maps due to map corruption, if the users are lucky to have non-corrupted maps; worlds like The Nether spiralling out of control with entities - with unmodified FTB setups running on Linux Centos.

Today it was FTB Ultimate 1.1.2 with crazy amount of entities on Nether/End with 8000+ entities approx. each and a small handful of chunks corrupted (4~10), FPS drops in certain locations (what I can only assume is the areas with these entity issues) what is causing this is beyond me at the moment, we reinstalled a fresh install (new unmodified FTB installation + new world) and again as soon as we got on it, same problem (we even tried an alternate machine to try and rule out any hardware issues to no effect).

http://www.minecraftforum.net/topic/302380-minecraft-region-fixer/ is being used to check maps. Maybe this will confirm the same problem for you.

Edit:
Another FTB Ultimate 1.1.2 server:
DIM-1: 7688 entities - 0 corrupted chunks - 27 region files.
DIM1: 1511 entities - 0 corrupted chunks - 7 region files.
DIM7: 3696 entities - 0 corrupted chunks - 17 region files.
DIM_MYST22: 39406 entities - 0 corrupted chunks - 78 region files.
...Stopped checking the worlds after the MYST22.

Confirmed 3 FTB Ultimate 1.1.2 servers at the moment have this issue, across 3 separate machines, each running Linux Centos, latest JAVA etc.

 

[dariusc93]

It likely stop because of the huge amount of entities. I would make a script to remove the entities or cut them down to a limit if the region fixer doesnt do it properly. If you have world guard installed do /stoplag. If you have clag or world edit install do /lagg clear or /remove items <radius> (or /butcher if its mobs). That will clear the entities up. Also delete the forechunks.dat incase something is loading chunks.

If your not using bukkit plugins, install and use aPerf and enable item grouping, and that should increase the tps.

 

[Draylynn]

After running Region Fixer I did actually install MCPC+ (to get plugins to load) on one of these servers, installed ClearLagg, entered the server did a killmobs command which killed a lot of mobs, however, very quickly the entity count continues to explode and it doesn't remove all the entities I believe/guessing because some of the effected map is not loaded due to lack of player presence. Some of these maps are quite big, simply running around mashing 'control+paste enter' isn't really viable.
RegionFixer can't seem to bring the entity count down 'at all' (I tried deleting the chunks, setting a new limit etc), I'm completely at a loss.

 

[dariusc93]

You would need to do /lagg chunk which should show you the area thats affected by the most entity count. also install world guard and do /stoplag which will delete all entities until you do /stoplag -c. You might want to mark the global region so mobs dont spawn so you can see if its mobs, or item entites.

 

[Draylynn]

That's one of the problems, /lagg chunk is showing for example:[ClearLag] Largest Chunk: x:23 z:-12 World:world with 8 entites!
Yet going to that chunk , digging it down to bedrock, it still reports the same entities and same chunk as 'largest chunk'.

Were testing /stoplag at the moment, so far 20% less RAM use, still watching the server, user is reporting a lot better performance.
As for the entities, it seems to be anything; Multiplying cows in swamps, caves with large numbers of creepers and various other nasty's in it, 15,000 items removed from a snow biome (didn't see any items anywhere) etc.

 

[dariusc93]

do /chunkloaders and check and see how many chunks are loaded. You may want to use NBT Explorer or a NBT->ASCII and check the forcedchunks.dat and locate any bloc that could be loading chunks, which could be the reason why the entities are still spawning. If that doesnt help, you may want to try to locate the chunk manually by setting up a script in Python or some language using the nbt libraries, have it scan each region, chunk by chunk and do a entity count where chunk X,Y is at.

For the ram part, add this to your startup script for the server

java -server -Xincgc -XX:NewRatio=3 -Xrs -XX:+UseThreadPriorities -XX:CMSFullGCsBeforeCompaction=1 -XX:SoftRefLRUPolicyMSPerMB=2048 -XX:+CMSParallelRemarkEnabled -XX:+UseParNewGC -XX:+DisableExplicitGC -Xnoclassgc -oss4M -ss4M -XX:+UseFastAccessorMethods -XX:+UseCMSInitiatingOccupancyOnly -XX:CMSInitiatingOccupancyFraction=75 -XX:+UseConcMarkSweepGC -XX:UseSSE=4 -XX:+UseCMSCompactAtFullCollection -Xcheck:jni -Xfuture -XX:+AggressiveOpts -Xmx4028M -Xms4028M -XX:+UseConcMarkSweepGC
-XX:+CMSIncrementalPacing -XX:ParallelGCThreads=[CORE_COUNT] -XX:+AggressiveOpts
-jar ftbserver.jar nogui

As i stated before, you dont want to run a server on the same machine as the client, so i would recommend getting a dedicated server soon.

 

[Draylynn]

We use some of those arguments on our servers, I've let our techies know about this thread and they'll play around with the additional ones.

/chunkloaders is great, except I have mentioned previously when a new map is created the issue is immediately present (there are no chunk loaders on map), installing essentials to get a count of the loaded chunks shows normal amounts loaded.
Forcedchunks.dat tidying up doesn't help... Editing them just gets really confusing. Ticket count 0, no chunks being forced. The server I'm playing with at the moment DIM-1 has 49000+ entities.
This entity issue appears to be happening at map generation/loading and persists. I have a feeling /stoplag will have to become routine (every hour or so it seems at the moment).

I work for the company sponsoring FTB, the server software is all on server racks tucked away in a data center, client is on my desktop/office computer. In fact for a majority of this, I don't need to login the client at all and run plugin commands via the console/SSH. So no worries there. I think this has been happening for a while on FTB in general, only recently it has severely escalated.
6 servers effected so far that have answered to my tweet this morning have all displayed the same problem, it's not even peak time yet... so it could get ugly for me later... *stocks up on coffee*, I think we need a mop.

Edit: 8 servers confirmed with this problem.

 

[Paul_T]

Hello,

Just a little more information, it'd be nice to see where these are coming from, because it suddenly started happening recently on a server that was fine, no changes on the server, just a migration from virtual server to physical hardware;

Scanning: 668 / 668 100% [######################################] Time: 00:23:53
 
Found 0 corrupted, 0 wrong located chunks and 1 chunks with too many entities of a total of 549984

It got significantly worse after the latest Ultimate update, we've been working closely with this customer for a while, but with this many entities (Which we can't find) we're really struggling to help him keep at 20TPS (With 60 - 80 players).