Player-count collection is profiler-independent. Population changes are recorded on player lifecycle events and reconciled once per minute. Hourly averages are time-weighted. Growth metrics use Player Administration first-seen/last-seen records, so new-player counts remain available across restarts.

Tracks Vintage Story's standard BlockEntity ... threw an error when ticked log entries by exact coordinate. It reuses the existing incremental log tail: no world scan, per-tick scanner, or extra Harmony patch is added. Destructive world controls are audited and require confirmation. Remove All Tracker Entries only clears saved tracker records; it never removes world blocks or block entities.

Recommendations combine active alerts and Smart Diagnosis. Clear dismisses the current recommendation until its condition disappears and occurs again. It does not fix the reported condition.

Clear removes retained alert history. Ongoing conditions can create new alerts; monitoring stays enabled.

Historical one-minute samples grouped by online player count. This is an observed workload model, not a hard player limit.

Aligned scheduler delay, CPU, players, loaded entities and managed heap. World-save markers include autosaves and manual saves; their overlap alone is not proof of a lag cause.

Execution intervals are listed first. Web elapsed and queue-wait counters include time waiting for the server; they are not CPU cost or proof of a cause. Callback wall times can include GC and scheduling pauses—inspect the spike evidence for overlapping measured work.

Built from the lightweight rolling diagnostics. Archived incidents retain approximately 30 seconds before, the incident itself, and 15 seconds after. Click a row for drill-down or select two incidents to compare recurring patterns.

Type-level growth compares the last two manual entity scans. Large increases are investigation leads, not automatic proof of a broken spawn rule.

The suite fingerprints the loaded mod/version environment. When it changes, it compares the previous environment's last known performance with the new one after enough diagnostics have collected.

Shows methods patched by more than one Harmony owner. Multiple owners can be completely compatible; this is a conflict-investigation lead, not automatic blame.

Compare Suite-owned queues and caches alongside heap readings. Refresh and export are manual; no heap walk or forced collection runs.

Schedule a one-time restart countdown without changing the recurring restart schedule. Example: restart in 10 minutes for maintenance.

Open this page to load server-main.log.

Server controls: Stop and Restart use the suite's clean shutdown scheduler with a 5-second safety delay. Restart still depends on Stratum or your external host/service/container supervisor to relaunch the process. Start is shown for control-panel symmetry but is disabled while this in-process dashboard is online; once Vintage Story stops, this dashboard stops with it and cannot start the process itself.

Input modes: Auto treats a leading / as a Vintage Story command and plain text as a server message. Command mode accepts commands with or without /. Message mode always broadcasts the text to online players. Input starts OFF and automatically expires after 15 minutes. No OS shell access is exposed.

Listener settings and public layers apply immediately. Terrain tile-size changes apply on the next export. Version 1.5.33 requires one Full Rebuild Map to repaint terrain with the corrected v11 hillshade; the existing v10 raw patch sidecars are still used for fast dirty-chunk updates afterward.

The one-shot checkbox is cleared after a successful export. Saving this card changes only export scheduling/mode settings and cannot toggle the public Web Map listener. When the map host is OFF, queued startup exports remain paused and live export controls are unavailable. Turning the map OFF also cancels an active live export after the current chunk load finishes.

No terrain status loaded.

Update Changed Terrain uses the persistent dirty-chunk journal and renders only dirty/new/missing tiles when the cache is compatible. Full Rebuild Map recreates the entire explored terrain cache and should be used after renderer/tile-size changes or for repair. Full rebuilds prefer parallel read-only save-database terrain workers; changed live chunks and compatibility fallbacks use the safe runtime world-loader path.

Administrator/Owner accounts can create, edit and remove any marker. Players create markers only in-game at their current position with /webmap waypoint add major|minor [icon:<type>] <name> and remove their own with /webmap waypoint remove <name>. Major and minor limits are enforced separately. Major markers remain visible farther out; minor markers appear closer in. Players can use /webmap live off|on|status to control whether their live player data appears publicly. Enter the same relative X/Y/Z shown by the in-game coordinate HUD; the public map translates them to its internal terrain origin automatically.

When the map runtime is OFF, terrain dirty-journal writes, map listeners, player/date tracking and export hooks remain inactive. When enabled, terrain still has no continuous scanner: block place/break events only mark their 32×32 map chunk dirty in a deduplicated journal. Administrators can queue a one-shot startup export or enable a startup-only interval such as weekly; neither option renders terrain while the server is already running. The public browser feed remains 30 seconds. Player coordinates remain at most 60 seconds. BlockLogger activity remains at most one query per 60 seconds while viewed.

2. Set the change

Generated changes are checked against loaded assets. Recipes apply after a server restart. Client item/block changes also need the managed content mod on clients; this generator does not hot-reload game assets.

Multiple Harmony owners on the same method is overlap evidence, not proof of an incompatibility. This scan is manual because reflection across every patched method is unnecessary during normal live updates.

Choose a target role above, then assign it to a player. Search saved players, including offline accounts. Direct privilege grants and denials are preserved. Discord role sync may reassign linked players later; review its mappings before removing a role.

Separates worker-queue delay, browser/network time, time waiting on Vintage Story's main thread, and actual main-thread work. This is the quickest way to prove whether the dashboard itself contributed to a lag event.

Every automatic collector should have a bounded cadence and avoid full world/entity walks. Manual heavy operations are clearly separated from background telemetry.

Circuit breakers disable only the failing nonessential feature. Core BlockLogger logging continues.

This answers “is the monitoring suite becoming part of the performance problem?” Background maintenance should normally remain tiny; manual scans are shown separately because they are intentionally heavier.

Workflow

A typical investigation is: Alert → incident black box → timeline → entity/chunk or memory evidence → manual engine profile when exact timing is required → measured optimization/config comparison → diagnostic bundle if support is needed.