Random "Read Timed Out" error

I run a Minecraft server for about 20-30 people, and lately it's been acting really weird. After a random amount of time (anywhere from 10 minutes for 48 hours) it just locks up. There is no other error from the console other than "Read timed out". The console locks up as well and doesn't respond to anything, pressing enter or typing anything just breaks the line and shows another ">". The CPU usage hit's 100% (out of 400%), and seems very steady while it's crashed. I've tried removing Spout, but that didn't work either. I'm currently in the process of removing plugin by plugin, but it is a public server and users get upset when you constantly have to restart, test, crash, restart. Nevermind their lack of resources after a plugin is removed. Is there any way or hope for fixing this? I started a new thread since other players seemed to have gotten another type of error or output before the "Read timed out" error, but mine just randomly happens.

The server never uses more than 10% of my RAM, but since the crashing, I've allotted the JVM 6GB on startup to possibly help it, which it didn't. It was always allowed to grow to 12G.

Here's a snippet of the server.log when it occurs.

Code:

2012-06-06 17:14:01 [INFO] [D+] [Architect] trophyguy21: yeah
2012-06-06 17:15:27 [INFO] Read timed out
2012-06-06 17:18:52 [INFO] Connection reset
2012-06-06 17:19:34 [INFO] Read timed out
2012-06-06 17:21:26 [INFO] Read timed out
2012-06-06 17:24:24 [INFO] Connection reset
2012-06-06 17:25:01 [INFO] Read timed out
2012-06-06 17:25:24 [INFO] Read timed out
2012-06-06 17:27:09 [INFO] Connection reset

Here are my plugins:

Code:

20:41:51 [INFO] Plugins (20): LazyRoad, WorldEdit, iConomy, CommandBook, OpenInv, LogBlockQuestioner, Vault, BanHammer, PermissionsEx, LogBlock, Votifier, Herochat, LWC, WorldGuard, dynmap, MagicCarpet, DynamicEconomy, VanishNoPacket, Superpickaxe, AntiCheat

Here is my server info:

Code:

2-06-06 20:38:36 [INFO] This server is running CraftBukkit version git-Bukkit-1.2.5-R3.0-6-gc598c4b-b2211jnks (MC: 1.2.5) (Implementing API version 1.2.5-R3.1-SNAPSHOT)

Here is my startup script:

Code:

java -server -Xms6G -Xmx12G -XX:PermSize=128m -XX:MaxPermSize=1024m -XX:+DisableExplicitGC -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:+UseNUMA -XX:+CMSParallelRemarkEnabled -XX:+UseAdaptiveGCBoundary -XX:-UseGCOverheadLimit -XX:+UseBiasedLocking -XX:SurvivorRatio=8 -XX:TargetSurvivorRatio=90 -XX:MaxTenuringThreshold=15 -XX:UseSSE=3  -XX:+UseFastAccessorMethods -XX:+UseStringCache -XX:+UseCompressedOops -XX:+OptimizeStringConcat -XX:+AggressiveOpts -jar craftbukkit.jar nogui

Here is my munin statistics: (You can tell when the server crashes when the CPU stays at a steady 100% utilization)
http://ipwnage.com/munin/
http://ipwnage.com/munin/localdomain/localhost.localdomain/cpu.html

I am running the server on Ubuntu 12.04LTS and Java 1.7, with 16GB ECC RAM, and a Xeon 3430. My internet speed is 100 up/down. The OS and minecraft is on an 80GB SSD, with backups every hour onto a 1TB drive. I have a cronjob that runs "save-all" every 10 minutes since the "Read timed out" errors would cause me to kill the java process, thus losing progress. This is all self hosted.


Any help would be greatly appreciated and I am willing to pay $10 to the person (or to each person) who helps solve this.

So just a quick update: It's still happening, but it seems that I can get the server running longer if I remove dynmap. Without, the server lasted about 24 hours before crashing, where, with it, it only lasted a few hours. When I restarted it after the crash, it crashed again almost instantly. It took me two more tries before it would run again. Could this mean it's a hardware issue or what?

Here's an image of the CPU usage; you can notice the crashes when it's at a constant 100% utilization.


EDIT by Moderator: merged posts, please use the edit button instead of double posting.

 

Wow, this must be the first post I've seen where I don't have to ask for more information.

If you have people on your server who are getting annoyed by the maintenance, get 'Modifyworld' (comes bundled with permissionsex) and set it to whitelist mode with a message like 'Server Maintenance - Try again later.'

Make sure your plugins are properly optimised for your server hardware (e.g. tell dynmap how many cores your server has, and set the rendering intervals so they don't slow down the server)

How big is your server.log file?

 

Heh, thanks. I try my best to get as much info as possible.

I would assume dynmap is only using one core; due to the configuration file:

Code:

# Parallel fullrender: if defined, number of concurrent threads used for fullrender or radiusrender
#  Note: setting this will result in much more intensive CPU use, some additional memory use.  Caution should be used when
#  setting this to equal or exceed the number of physical cores on the system.
#parallelrendercnt: 4

Should I uncomment that and set is to 2? I have been trying to fullrender, but am never able to complete it because it keeps crashing, and it doesn't save it's queue, which causes me to restart the entire render.

Also, this is in the file:

Code:

# How many update tiles to work on at once (if not defined, default is 1/2 the number of cores)
tiles-rendered-at-once: 2

My server.log is about 5MB, I have cronjob that archives it after each day.

Don't mean to double post, but I think I fixed the problem. I've made dynmap to parallelrendercnt: 3. And so far everything is running very smooth; without any problems.

*knocks on wood*

EDIT: It lasted a whole two days, I'm shocked. It's back to being all derpy again.

EDIT by Moderator: merged posts, please use the edit button instead of double posting.

 

Um I have the same problem but I don't have Dynamic Map. I also have a Mac so this is kind of confusing. I have my suspicions though that it has to do with connection problems because ever since I got log block my server keeps on doing
read timed out.
I don't want to get rid of log block because that is essential to my server but I don't know what to do. I believe that the my sql part of it is slowing my connection down but sometimes when certain players log on it stops. I really have no clue what is happening.

 

Yeah, i have the same problem, only with 1.3.1. It started when i made the upgrade from 1.2.5 to 1.3.1. Now my server randomly displays the message "read timed out" and when a player leaves, the server doesn't say anything at all. If i do /list it will appear as though the player never left. When the same player joins the server once again, a clone is created of them and stays on the server. If i do /list, it displays the player's name several times.