Diagnosing Can't Keep up mesages

[jghake]

Title:

Diagnosing Can't Keep up mesages​

Launcher Version:

1.4.8​

Modpack:

FTB Infinity​

Modpack Version:

1.10.1​

Log Link:

​

Details of the issue:

I am hosting a server for me and 4 friends and we are getting a lot of can't keep up messages. I ran opis when we first started having very bad problems and it pointed to two of my buddies "factory" filled to the brim with over 3000 BC pipes with massive amounts of diamond sorting pipes.

They have since removed that setup and went with an AE system and the server is running much less laggy but the can't keep up messages still persist.

I am running the server on MineOS, in a Hyper-V VM, on an HP Proliant ML350 G5.

8 cores exposed to the VM (I have dual quad Xeon's)
The VM's virtual disks are stored on a RAID 5 volume with 3 7200 RPM sata drives.
MineOS is allocated 15 GB of RAM, the FTB server has min 4 max 12.
I have a gigabit internal network, and 120/12 WAN.

Here are my java args:
-XX:UseSSE=4 -XXermSize=256m -XX:MaxPermSize=512m -XX:-DisableExplicitGC -XX:+UseParallelOldGC -XXarallelGCThreads=4 -nogui

No changes to the modpack other than we added Dynmap. I've moved the Dynmap directory out of the server directory to exclude it from the snapshots and backups. I was noticing VERY large files before I did this.

We used to only get the can't keep up messages on the top of every hour when MineOS would take a backup/snapshot but now we get them very regularly.

Here are a few.

[13:13:29] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 4550ms behind, skipping 91 tick(s)
[13:13:18] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 9036ms behind, skipping 180 tick(s)
[13:12:56] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 6259ms behind, skipping 125 tick(s)
[13:12:39] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 6245ms behind, skipping 124 tick(s)
[13:12:21] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 5875ms behind, skipping 117 tick(s)
[13:12:08] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 8026ms behind, skipping 160 tick(s)
[13:11:48] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 6638ms behind, skipping 132 tick(s)
[13:11:30] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 8017ms behind, skipping 160 tick(s)
[13:11:08] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 5203ms behind, skipping 104 tick(s)
[13:11:00] [Server thread/WARN]: Can't keep up! Did the system time change, or is the server overloaded? Running 12093ms behind, skipping 241 tick(s)

Right now there are 2 guys on, memory usage is over 10GB of the allotted 12 maximum.

Where should I take a look?​

 

[Henry Link]

Typically can't keep up messages are related to TPS of the server and the CPU speed. So I noticed an overall TPS increase on my server when I started using java 8 64bit instead of java 7. Also, cut back the RAM assigned to java/minecraft. Having to much actually increases garbage and heap processing and slows java down. Since it is only for the five of you use about 4 GB of RAM instead. Increase only if it is needed.

What is the actual CPU speed of your server?

Also, keep in mind that as time goes on and you add more modded blocks/structures (AE2, BuildCraft, etc.) that everything does add more processing time to the loop and can cause TPS issues as well. Sometimes it may also be an issue or bug with the way somebody set something up (back stuffed pipes, AE2 crafting, Steve's Factory Manager programming, etc.).

 

[jghake]

Thanks for such a quick responce! I've decreased the RAM. In the UI for MineOS it is now showing its using 5000 of 4096 Mb of RAM. Is this going to be an issue?

My Java version is 7u79-2.5.6-1~deb7u1

I have 2 Xeon E5410's. 2.33 GHZ, 12Mb cache, 1333 Mhz FSB.

http://ark.intel.com/products/33080/Intel-Xeon-Processor-E5410-12M-Cache-2_33-GHz-1333-MHz-FSB

 

[Henry Link]

To be honest the CPU might be a little lacking. 2.33 GHz isn't the best for running modded minecraft. But try to update the java to version 8 and remove the PermSize commands as that isn't used with Java 8. I'm hopping all of this helps but the I've seen my server TPS lag with can't keep up messages at times and it is quad xeon's running 3.4 GHz.

 

[jghake]

I upgraded both Xeon's to 3.16 GHz for less than $40. I don't get nearly as many alerts and tps is much better.

Nobody is on right now but here is what we have.

[09:25:49] [Server thread/INFO]: Lava Age [6]: 20.00 TPS/0.52MS (100%)
[09:25:49] [Server thread/INFO]: Mining World [8]: 20.00 TPS/0.40MS (100%)
[09:25:49] [Server thread/INFO]: The Last Millenium [-112]: 20.00 TPS/0.04MS (100%)
[09:25:49] [Server thread/INFO]: Dungeon [-34]: 20.00 TPS/0.08MS (100%)
[09:25:49] [Server thread/INFO]: Nether [-1]: 20.00 TPS/1.84MS (100%)
[09:25:49] [Server thread/INFO]: Overworld [0]: 20.00 TPS/7.25MS (100%)
[09:25:49] [Server thread/INFO]: Overall: 20.00 TPS/10.20MS (100%)

It would also seem that the RAM usage shown by MineOS is OK too. Apparently they are just pulling the memory usage of all Java threads and combining them into the one display. Things are running MUCH better.

I upgraded to Oracle Java build 1.8.0_60-b27 and removed the permsize commands.

Hopefully we are good to go!