Skip to content

What NOT to do!

Because common sense is not common, this page collects the things that go wrong with this plugin. They fall into two very different buckets, so they are split below:

Read the config files first

PLEASE read the configuration files in your RTP folder (also visible in this repository's resource folder) before asking for setup support. A lot of information is easily accessed and well described within the configuration files.


This will break the mod entirely

These mistakes prevent teleportation from working correctly.

Don't blacklist every biome (or whitelist zero biomes)

If no biome is eligible, every teleport attempt has nowhere valid to land, so teleportation fails outright. Always leave at least one usable biome.

Don't run conflicting /wild or /rtp providers together

Mixing plugins that both register /wild or /rtp without resolving the command conflict means the wrong plugin (or no plugin) may answer the command. Resolve the command conflict before running both.

Don't delete the default region/world configuration

LeafRTP uses the default region and world entries as the template for any new world or region. Removing them leaves the plugin with nothing to build from, and it will not behave correctly. See Intended usage.

Don't change definitions/regions/default.yml to world_nether just to teleport in the Nether

Changing world: "world_nether" on the default region redirects all unconfigured worlds (including your main overworld) into the Nether. Instead, create a dedicated region file (definitions/regions/nether.yml with world: "world_nether", vert.maxY: 120, and vert.requireSkyLight: false) and map definitions/worlds/world_nether.yml to region: "nether" via /rtp editor, /rtp admin setup, or /rtp config worlds world_nether region=nether.

Don't rely on worldBorderOverride: true when shape.radius with unit suffixes will do

worldBorderOverride: true is generally not recommended because it replaces your entire shape: block with a plain square and discards your shape.name, shape.radius, and inner spawn-exclusion hole (shape.centerRadius). Instead, set shape.radius in /rtp editor or definitions/regions/<region>.yml using an explicit distance unit suffix (such as 2500b or 5km).


Why is this running slowly?

These setups work, but they make teleports expensive - either lagging the player who teleports or stalling the whole server.

Don't grant rtp.unqueued (or wildcard rtp.*) to regular players

Out of the box, LeafRTP pre-generates and verifies safe locations asynchronously in the background (cacheCap in definitions/regions/default.yml), so /rtp pops a ready spot without main-thread or chunk-generation lag. Granting rtp.unqueued (or rtp.*) to default players forces every /rtp to bypass the pre-generated cache and run live on-demand chunk selection and loading, which causes server lag spikes when multiple players teleport. Keep default players on rtp.use only.

Don't pass dynamic parameter overrides to regular players

Passing dynamic override parameters (shape=, radius=, centerX=, centerZ=, etc. under the rtp.params permission) constructs an ephemeral, throwaway region on the fly. Because temporary regions are uncached and have no persistent spatial memory, every invocation forces on-demand chunk loading and real-time safety verification. Keep dynamic parameters restricted to staff testing, and configure permanent named regions for everyday player use.

Don't wire override parameters into player commands, signs, or portals

Override parameters (shape=, radius=, centerX=, ...) build a throwaway temporary region that is uncached and remembers no past failures, so every call pays full generation cost and can stall under load. Use them only to test settings, then bake them into a real named region and point players at a plain /rtp. (Targeting an existing region with world=<name> or region=<name> is fine - those stay cached.) See Intended usage.

Don't enable syncLoading: true or raise viewDistanceSelect in advanced/performance.yml

Setting syncLoading: true forces chunk loading onto the main server thread (keep syncLoading: false), and setting viewDistanceSelect above 0 loads surrounding chunks on every candidate check during search (keep viewDistanceSelect: 0). Because autoTune: true scales background compute relative to your configured baselines rather than overriding them, these settings compose together: tightening syncAllottedTime (e.g. 2-5 ms) or maxMSPT (35.0-40.0 ms) with autoTune: true tightens the backend tick budget (invisible to players while the hot queue stays populated), while viewDistanceSelect: 0 (surrounding chunks load after arrival) and raising pregeneratedPreference toward 0.8-1.0 (biasing toward already-generated terrain on disk) reduce per-candidate I/O work.

Don't over-restrict biomes

Blacklisting most biomes (or whitelisting only a few) leaves so few valid landing spots that the engine has to reject many candidates before finding one, which causes major lag during teleportation even when it eventually succeeds. (Restrict all of them and you cross into the "breaks entirely" bucket above.)


See also