bergerkiller / NoLagg

Prevents common lag causes on your server
27 stars 14 forks source link

Random Bukkit Crash #13

Open sociableWeaver opened 12 years ago

sociableWeaver commented 12 years ago

13:33:14 [INFO] This server is running CraftBukkit version git-Bukkit-1.1-R3-b1846jnks (MC: 1.1) (Implementing API version 1.1-R3)

13:33:38 [INFO] Plugins: Minequery, WorldEdit, Suicide, WebChat, CommandBook, ColoredSigns, ClearInv, Help, AutoGamemode, HeroChat, FactionsSS, LogBlockQuestioner, PermissionsEx, WorldGuard, BorderGuard, ExactSpawn, HeroicDeath, MCSL, Vanish2, pvparena, SpamHammer, Permissions, FirstJoin, KiwiAdmin, NoCheat, TelePlusPlus, Votifier, MobDisguise, MobArena, AutoAnnouncer, LogBlock

NoLagg worked wonders on our server. But sadly we have had to disable. Our server would stop responding randomly. There was nothing in the logs from bukkit or your plugin that are of any help. If we caught the crash quickly the stop command would work in console. If we didn't catch it for an hour or so we could type the stop command but nothing would happen. We would end up having to kill the java process to terminate the server. After disabling your plugin this random crash ceased. I'm sorry I don't have more information on it. Feel free to ask me more questions and I will answer them to the best of my ability.

Note: FactionsSS is a custom plugin that does not affect the mechanics of MC in anyway.

NINJ4 commented 12 years ago

Our config.yml:

#Sets if and how many items are buffered per chunk
bufferItems: true
maxItemsPerChunk: 40
bufferTNT: true
tntDetonationInterval: 1
tntDetonationRate: 10
explosionRate: 5
useSpawnLimits: false
spawnlimits:
  default:
    mob: 1000
useChunkUnloadDelay: true
chunkUnloadDelay: 10000
formItemStacks: true
stackRadius: 1.0
stackThreshold: 2
ignoredItemTypes:
- DIAMOND_PICKAXE
autoSaveInterval: 400
updateInterval: 20
monitorInterval: 40
autoFixLighting: true
useBufferedChunkLoading: true
chunkSending:
  enabled: true
  minRate: 0.25
  maxRate: 2.0
  triggerRate: 0.5
  sendThreadCount: 2

Our log.txt:

NoLagg enabled: 2012.02.09 18:31:06
NoLagg disabled: 2012.02.09 18:35:42
NoLagg enabled: 2012.02.09 18:38:54
NoLagg disabled: 2012.02.09 22:06:24
NoLagg enabled: 2012.02.09 22:06:29
NoLagg disabled: 2012.02.10 15:04:31
NoLagg enabled: 2012.02.10 15:05:44
NoLagg disabled: 2012.02.10 22:24:46
NoLagg enabled: 2012.02.10 22:25:27
NoLagg disabled: 2012.02.11 7:30:53
NoLagg enabled: 2012.02.11 7:31:11
NoLagg disabled: 2012.02.11 13:19:52
NoLagg enabled: 2012.02.11 13:20:03
NoLagg disabled: 2012.02.11 13:28:35
sociableWeaver commented 12 years ago

We will turn it back on and run for several hours with the logging set to high.

NINJ4 commented 12 years ago

We enabled NoLagg at 2012.02.13 14:15:57, and had to disable it and restart the server instance approximately 2 hours later at 2012.02.13 16:28:36 due to complete irresponsiveness of the server to join attempts. Where the player count seems to remain at an eerily constant 17 throughout the last 30 minutes or so of this log, during which time no player was able to join and the players listed as joined were in fact disconnected, is believed to be the period of the hang. The server was brought to a stop from console.

Without NoLagg installed, we are able to keep the server instance running almost indefinitely, but as you can see, with NoLagg installed, the server becomes unresponsive after only a few hours. We hope you can look into this, as this plugin greatly improved our server's responsiveness.

Here is the relevant log: http://pastebin.com/G5HpC5bZ

bergerkiller commented 12 years ago

All features reside in separate jar files now. Removing the main NoLagg.jar will most likely fix it, as it contains a lot of broken, old or slow code. (I am moving the healthy bits out, fixing them and testing them thoroughly.)

May you have some issues still, feel free to test what component is causing it. You can use the /lag examine command to see what plugin is currently locking the server, if possible it will find out what is causing it right away. (if it really lags, use a small timespan, such as /lag examine 50)