Skip to content

AtlasSpigot 26.2 - build 17 (maximum tuning)

Choose a tag to compare

View all tags
@AtlasStudioMC AtlasStudioMC released this 05 Sep 15:41
atlasspigot-26.2-17
8cd8083

Config update for Minecraft 26.2. The jar is unchanged — this is the third and final tuning pass.

⚠️ This build can break a plugin

Builds 15 and 16 changed gameplay. This one crosses into plugin compatibility.

hopper.disable-move-event is now on. InventoryMoveItemEvent no longer fires. Plugins that watch hopper transfers — shops, item sorters, some economy plugins — will stop seeing them.

It is the single largest performance win in the entire config, which is why it's here. It is also one line to undo: set hopper.disable-move-event: false in config/paper-world-defaults.yml and those plugins work again, at the cost of the win.

If you run a shop or sorting plugin and don't want to think about it, build 16 is identical except for this and remains available.

Also in this pass

  • bukkit.yml is now shipped — it wasn't before, so spawn limits were running at stock. Monsters 70→40, animals 10→6, ambient 15→4, water-ambient 20→8, spawn attempts halved. Every spawned mob costs ticks for as long as it lives, so this is one of the biggest wins available. Caves feel emptier and farm rates drop.
  • sleeping-block-entity — Lithium's optimisation; idle block entities stop ticking until poked.
  • throttle-mob-spawning — chunks that fail to spawn mobs 8 times get a 25% chance thereafter, cutting wasted attempts in lit areas.
  • hopper.ignore-occluding-blocks — hoppers skip occlusion checks on pickup.

Cumulative warning

Stacked with builds 15 and 16, this configuration is tuned hard and is not vanilla-feeling. Spawner mobs have no AI, monsters despawn at 28/48, mob caps are cut by ~43%, simulation distance is 3, and entities are broadcast at half range. If you want a gentler server, earlier builds are still listed.

Verified

Boots clean in 7.16s through start.sh. spigot.yml, atlas-global.yml and paper-world-defaults.yml all return byte-identical after the server writes them back, and every changed value was confirmed still in place.

Not benchmarked. No measurements taken on 26.2, so no figures are claimed. Every setting, including what stays off and why, is in NOTES.md.

Assets 4
Loading