search

Zrips / CMI

116 stars 97 forks source link

Huge workload at CMI! #8410

Open imtorich opened 8 months ago

imtorich commented 8 months ago

Is this even a premium plugin or what did I give my money for? I have 90% of plugins turned off, everything is configured perfectly, and a huge load on server performance with online more than 60! Please resolve this, we don't want to post a bad review and change the plugin. Please help us.

Our spark report was generated with an online of 60 players, 23 minutes of analysis, with a filter if over 150 ticks. https://spark.lucko.me/eLFBX7xXfs

imtorich commented 8 months ago 
[19:24:20 INFO]: PESENkASPETA issued server command: /rtp
[19:24:20 INFO]: --------------------------------------------------
[19:24:20 INFO]: CMI plugin version: 9.6.9.8 SqLite-> 9.6.9.9.
[19:24:20 INFO]: CMILib: 1.4.4.6 -> 1.4.4.7
[19:24:20 INFO]: Server version: Plazma 1.20.1-R0.1-SNAPSHOT-.
[19:24:20 INFO]: CMI Economy: On Vault Version: 1.7.3-CMICMI Chat: Off 
[19:24:20 INFO]: Modules -> 20 enabled 39 disabled: spawnerProximity, cuffed, customMessages, mirror, tablist, armorEffects, moneyCheque, skin, elytraBoost, rideAnimation, elytraLaunch, bossBarCompass, coloredArmor, chatBubble, votifier, shulkerBackpack, cooldown, headDrop, attachedCommands, spawnerCharge, interactiveCommands, namePlates, timedCommands, signEdit, hpBossBar, holograms, anvilRenameColor, jail, paintingEditor, dynamicSigns, playerChatTag, durabilityLoss, deathMessages, disabledEnchants, worldLimitsElytra, ranks, firstJoinMessages, silentChest, warmup
[19:24:20 INFO]: --------------------------------------------------
Zrips commented 8 months ago

From your log it just looks like CMI had some extra work with /cmi status command, you might have had quite a few chunks loaded in and CMI processed all the data, it should not really affect servers work as of your timings have been 20tps across the board

imtorich commented 8 months ago

From your log it just looks like CMI had some extra work with /cmi status command, you might have had quite a few chunks loaded in and CMI processed all the data, it should not really affect servers work as of your timings have been 20tps across the board

spark reports a TPS of 20 (good), minimum MSPT ~ 35.5 ms (good!), But maximum MSPT ~ 116 ms (over 50 ms, that's bad!). If I run twice as many players on the server, the server will be very bad! And yes, the status command was not used in the measurement. Also, I'm wondering why all players need to be stored in memory?

[19:51:30 INFO]: Platform: amd64 (Linux), Running threads: 203 [19:51:30 INFO]: TPS: 20.00 (19.99 19.99 19.98 19.99) [19:51:30 INFO]: CPU utilization: 21.23% (16 cores) [19:51:30 INFO]: Uptime: 15 hours 50 minutes 49 seconds. [19:51:30 INFO]: Memory Utilization: 44.31% (5444.00/12288 MB) [19:51:30 INFO]: Java Version: 21.0.2+13-LTS-58() Build: [19:51:30 INFO]: Disk Usage: 32.49% (297/914 GB) [19:51:30 INFO]: 1. world 13840 chunks, 5527 entities, 10915 tiles, 53 players [19:51:30 INFO]: 2. world_nether 176 chunks, 30 entities, 1 tails, 1 players [19:51:30 PM INFO]: 3. world_the_end 49 chunks, 0 entities, 20 tiles, 0 players 19:51:30:30 INFO: 4. spawn 419 chunks, 11 entities, 63 tiles, 6 players

imtorich commented 8 months ago

Exactly when doing the report yes, used the status, but before that there was also a heavy load, 400 MSPT sharp. Here is the report of the previous sparc, without status. https://spark.lucko.me/W4OARqtRmn