(General Help) Java lock ups, read timeouts.

Discussion in 'Bukkit Help' started by MikeA, Nov 8, 2012.

Thread Status:
Not open for further replies.
  1. Offline

    MikeA

    1. What OS are you using? CentOS 5.8
    2. What architecture is the OS you are running (x64 or x86)? 64-bit (x64)
    3. What version and architecture is your java install? Oracle Java 7 JRE 1.7.0_07-b10
    4. Are you running any wrappers? Remote Toolkit (latest version)
    5. What build of CraftBukkit are you running? 2455
    ---
    6. What command are you using to run your CraftBukkit server? java -Xmx10G -Xms10G -jar bukkit.jar
    7. What plugins are you running? http://pastebin.com/cD7k4LL0 (Most, removed a few custom ones)
    8. What error are you getting? "Read timed out" -> "Connection reset"


    This help request is for general java and server use, not Craftbukkit. I run two servers, Tekkit 3.1.2 and Craftbukkit 1.4.2, every 15 minutes or so both of our servers (in separate screen sessions) crash with "Read Timed Out" errors and everybody disconnects. There are no other errors and after about 15 seconds it is fine again but everybody disconnects.

    I can only assume this is a problem with Java itsself since BOTH instances of java (bukkit + tekkit) are crashing near or at the same time. CPU usage goes to zero basically during this time. Don't say "Lock, we don't help with Tekkit". This is not a "Tekkit help thread", it's Java, this is mainly happening on my Craftbukkit server.

    If anyone has any clues what it could be, please suggest.
     
  2. Offline

    LaxWasHere

    My guess is you're being DDOSd.
     
  3. Offline

    MikeA

    I have checked all network adapters and monitored them in real-time and there is no attack. Also my bandwidth graphs show nothing out of the norm.

    After the servers 'un-freeze' my outbound bandwidth spikes up to 10-30 Mbit though, so it seems like the connections/packets are being "froze" or something for a few seconds then released. (Yes, stupid way to explain, but hell..)
     
  4. Offline

    TnT

    Read time outs are caused by a plugin. What similar plugins do you have betweem both your servers? Do you use the same mysql instance for both? Are any of your plugins writing to the same database?

    Did you remove any plugins in an attempt to eliminate these issues?
     
  5. Offline

    cnaude

    With a 10GB heap you might be running into long pauses caused by the JVM gabage collector. You might want to try running the JVM with an incremental garbage collector and see if that helps. If you have access to jvisualvm or jconsole you should be able to watch what is happening in real time.

    Code:
    java -Xincgc -Xmx10G -Xms10G -jar bukkit.jar
    
     
Thread Status:
Not open for further replies.

Share This Page