-
Notifications
You must be signed in to change notification settings - Fork 10
Diagnostics
Leaf26 edited this page Jun 17, 2026
·
1 revision
/rtp info is the operator diagnostics dashboard. It reports loaded worlds, permanent regions, queue statistics, and learned-state (spatial memory) summaries - the live view of whether the engine is keeping up. The charts and counters render from in-memory state, so reading them never touches the teleport hot path.
/rtp info # whole-server overview (permission: rtp.info)
/rtp info region=<name> # focus a single region
/rtp info world=<name> # focus a single world
| Group | What it tells you |
|---|---|
| Worlds & regions | Which worlds are loaded and which permanent regions exist. |
| Cache depth | Ready-to-serve locations vs capacity (cached / keptCache / unkeptCache / cacheCap). A healthy region keeps this near its cap. |
| Backlog | Unverified backlog buffer depth (backlogCache / backlogCacheCap). |
| Queue & in-flight | Players waiting on a coordinate (locationQueue) and calculations in progress (inFlightCalculations). |
| Pipeline latency | Teleport-pipeline percentiles (pipelineMsP50 ... pipelineMsP99) and the slow-pipeline count over the configured threshold. |
| Spatial memory | Coverage and bad-location stats (memCoveragePct, memBadPct, memBadCount) and the top rejection cause (memTopCause / memTopCausePct). |
| TPS / MSPT | Server tick health; on Folia, a per-region breakdown. |
-
Cache sitting near zero with a non-zero queue means generation is not keeping up with demand. Run
/rtp scanto build spatial memory, and review performance.yml (attempt limits, cache rate,cacheCap). -
High
memBadPct/ a dominantmemTopCausetells you why candidates are being rejected (biome, unsafe block, claim). If a biome or block filter is too aggressive, adjust safety.yml or the region's biome controls. - Rising pipeline percentiles point at chunk-load or verification cost; confirm the Anvil pre-filter is doing its job and that spatial memory is populated.
-
Low TPS / high MSPT during
/rtpbursts is the symptom the engine is designed to avoid; if you see it, capture/rtp infooutput and a timings report for a bug report.
Every counter above is also available as a PlaceholderAPI placeholder (%rtp_<key>%) and inside messages.yml as [key], so you can surface queue depth or latency on a scoreboard or admin panel without /rtp info. The metrics snapshot knobs (TPS/MSPT/heap aggregation, mostly Folia) live in metrics.yml.
See also: Performance, Metrics, PlaceholderAPI, Scan and Spatial Memory.