Need help optimizing server for high player counts

Discussion in 'Spigot Help' started by d1m0a1n, Jun 3, 2017.

  1. Recently, my server has been getting more players. At peak times, there is usually ~120-140 players online. At these times, the TPS can dip pretty low, to 6-10. At player counts around or below 100, the server runs fine with 19.5-20TPS. I've changed some of the server configs based on the recommended optimizations here, as well as using different java flags, which can be found in the timings report below.

    The timings don't really show any specific plugin causing lag, but general gameplay aspects, like mobs and some redstone. I have plugins like MobStacker to reduce entities, there's usually less than 3000 in the main world.

    Server Specs:
    RAM
    4 x 8192 MB

    Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
    Number :
    8
    Cache :
    8192 KB
    Speed :
    4200 MHz
    Disks
    1 x 240GB SSD

    Kernel version
    "Ubuntu 14.04.4 LTS" 3.14.32-xxxx-grs-ipv6-64

    The server is running through BungeeCord, on Paper Spigot git-Paper-1104


    Short timings report.


    Any tips for optimizations are appreciated.
     
  2. Gaxan

    Supporter

    You're spawn is a lag hole.
    Everyone pointed this out the last time you asked for help and you ignored everything.

    You have 3,500 entities and 3,500 tile entities mostly in one area. Until you get rid of them or put your worlds on individual servers it's going to keep lagging.
     
  3. Last time I asked for help was an entirely different issue, which was resolved.
     
  4. Gaxan

    Supporter

    I'll try to explain. Every time someone joins spawn, it's using excessive resources. The more people you have online the more it's going to lag with all of the signs and entities you have in that area. The CPU is only going to handle so much before it's maxed out. Since you are running a dedicated, clone the current server, add a hub and set a player limit. If one is full they can connect to the other. This will allow you to use multiple cores on your dedicated.