Monitoring Dashboard — Monitoring

Accessing the Dashboard

In the dashboard's side menu, open Monitoring. Monitored servers lists each server with its status, TPS, MSPT, players, open alerts and TPS over the last 24 hours. Click a server, or Open, to reach its dashboard; the open-alerts badge opens it on the Alerts tab. From Servers, the pulse icon on a server card (Live monitoring) leads to the same page.

A server's dashboard has four tabs:

Monitoring: the indicators and charts described on this page;
Alerts: the server's active alerts, its rules and their history (see Metric Alert Rules and Event Alert Rules);
Profiling and Memory: the performance profiles and memory reports the server uploaded. These two tabs are there even without a plan that includes monitoring.

The Maintenance button at the top silences the server's alerts for a while (see Maintenance and Mute).

KPI Cards

At the top of the dashboard, you'll find quick-glance performance indicators:

TPS — Current ticks per second (green ≥19, yellow ≥17, red <17)
MSPT — Current milliseconds per tick (green ≤50 ms, yellow ≤100 ms, red >100 ms)
Players — Current player count, with the peak of the displayed period
RAM — Current memory use in GB, with the peak (green up to 70 % of the maximum, yellow up to 85 %, red beyond)
CPU — Current CPU use of the server process, with the cores used, the system CPU and the peak (green up to 60 %, yellow up to 80 %, red beyond)
Uptime — Real uptime duration when online, with restart count and the availability of the displayed period: the share of time covered by data, counted from the first data ever received and over closed periods only

Colours apply to a live value only: the server is online and its last snapshot is less than 15 minutes old. Otherwise the value is grey, with Last measurement … ago.

Below the cards, server health goes past the average TPS: the slowest tick, the 95th percentile of tick time, ticks over the 50 ms budget, time lost behind real time and garbage collection pauses; memory left after garbage collection against the maximum heap, and the whole process against the container limit (an out-of-memory kill looks like an unexplained crash), summed up as Headroom OK, Getting tight or Out of memory soon; and the busiest worlds. These need a plugin version that measures them. On Folia, a Folia regions panel details each region.

The status badge also says when a server that still sends heartbeats is Frozen (no tick for 10 seconds) or Paused (empty server) because Minecraft suspends an empty server, and tells a Stopped server (announced stop) or a Monitoring paused one apart from a server that went quiet without notice.

Charts

Six dedicated charts in the Split view:

TPS & MSPT

Dual Y-axis chart showing TPS (left, 0-20 scale) and MSPT (right). For aggregated ranges, a light band shows the min–max range of each. On Folia, the lines carry the worst region, with TPS, region average and MSPT, region average alongside.

Players & Entities

Online players and loaded entities, as two areas on a shared axis (not stacked).

RAM Usage

Used RAM against Max RAM (dashed line), in GB.

CPU Usage

Multi-layer chart:

System CPU (grey dashed) — total machine load
Server CPU (orange) — the Minecraft server process
Cores used (purple line, right axis) — equivalent cores consumed (server CPU × cores / 100)
CPU peak (shaded area, aggregated ranges) — the highest value of each interval

Loaded Chunks

Area chart of loaded chunk count.

GC Pauses

Dual-axis chart:

GC pause (pink area, left axis) — pause time in ms per snapshot; the average per interval on aggregated ranges
Collections (purple line, right axis) — number of collections per snapshot; the total per interval on aggregated ranges

Every chart shares the same cursor, shades the periods without data, and draws the thresholds of your metric alert rules as red dashed lines. Times are shown in your browser's time zone.

Time Ranges

Preset Ranges

Click a preset button to view data for the last 15m, 30m, 1h, 6h, 12h, 24h, 2d, 7d, 14d, or 30d.

Available ranges depend on your plan:

Pro: up to 7 days
Enterprise and hosting provider accounts: up to 30 days

A range beyond your plan shows a lock (Beyond your plan: up to …).

When a preset range is selected, use the ← → arrows to navigate backward and forward through time:

← shifts the window back (e.g., from now-30m→now to now-60m→now-30m), as far as your plan keeps data
→ shifts forward (disabled when already at "now")
Now jumps back to the current time

The exact time window is displayed when navigating history.

Custom Range

Click Custom to pick a start and an end (From, To, then Apply). The window can be neither wider nor older than what your plan keeps.

Previous Period

Compare with previous period adds the previous window of the same length to the charts, so that you can see what changed.

View Modes

Two chart layouts, from the toolbar's View menu:

Split — One chart per metric (default)
Overlay — The selected metrics on a single chart, Combined Metrics. Each one is drawn on its own observed range (0 % = its lowest, 100 % = its highest in the window); hover for the real values

In Overlay, Select metrics chooses among TPS, MSPT, players, entities, used RAM, server CPU, loaded chunks and GC pause (TPS, MSPT, CPU and RAM by default).

Event Markers

When events occur (server starts and stops, alerts, maintenance, moderation actions, etc.), they appear as vertical dashed lines on all charts. Each line has a small icon at the top, coloured by severity (blue for info, amber for warning, red for error):

a cross for lifecycle events (start, stop, alerts, maintenance)
a pulse for performance events
a shield for moderation and security events
a wrench for custom events

A benchmark run is drawn as a shaded purple zone labelled Benchmark. The Show on charts menu chooses which categories are drawn (all by default), and also shows the profiles and memory reports the server uploaded, marked P and M.

Hover near an event marker to see its title in the chart tooltip.

Auto-Refresh

The dashboard auto-refreshes data every 15, 30 (default) or 60 seconds, or not at all. Auto-refresh pauses automatically when navigating historical data.

Export

The toolbar's Export menu offers:

Export CSV — The chart data as comma-separated values
Export JSON — The chart data as structured JSON
Export events (CSV) — The events of the window

Each chart card also has its own button to export that chart as a PNG image.

Fullscreen

Click the maximize icon on any chart card to expand it to fullscreen. Press Escape or click the close button to exit.

Events Table

Below the charts, the Benchmarks card gives the regression verdict of a verified server and the benchmarks run during the window. Then the Events panel shows a chronological table of the events in the selected time range, up to 1,000. Filter by category or source using the dropdown menus (All categories, All sources). Uploaded profiles and memory reports appear in it as Capture lines.

Columns: Severity, Title, Category, Source, Type, Time.

Public Status Page

Give players a page that answers one question: is the server up? It is off by default. Turn it on from Dashboard → Servers → Manage → Monitoring tab → Public status page & external probe, and choose its Address: voxelbench.com/status/your-address (3 to 40 lowercase letters, digits or single hyphens, neither first nor last; a few words are reserved, and an address can be used by one server only).

The page shows:

the server name and icon;
its status: Up and running (an empty server paused by Minecraft included), Not responding (online but the main thread is frozen), Under maintenance (a maintenance you started), Down (an announced stop, or a server that went silent), or Status unknown (monitoring paused, off, or not yet reporting);
its daily availability over the last 30 complete days, from the same daily figures as your dashboard;
when it started, stopped, or went into and out of maintenance.

It never shows measurements (TPS, players, memory…), player names, maintenance reasons or event details. It asks search engines not to index it: share it by link. It refreshes every minute. Closing it takes it offline within a minute; the address is kept, so reopening it gives the same link.

External Probe

To watch the server with Uptime Kuma, BetterStack or any uptime tool, click Create a probe URL in the same panel. The URL is shown once: copy it then, or Regenerate it later (the old one stops working immediately).

GET https://voxelbench.com/api/v1/monitoring/health/vbh_…

200 while the server is running (or paused by Minecraft because it is empty), 503 otherwise (down, not responding, monitoring paused or off). HEAD works too.
Body: {"status":"up"|"down","state":"running"|…,"maintenance":true|false,"checked_at":"…"}.
During a maintenance you started, the code stays honest (503 if the server is really stopped): pause your uptime tool for the window, or match on "maintenance":true.
It does not count against your daily API quota. It is limited to 12 checks per minute per server; one check a minute is plenty.
The URL names no server. Treat it as a secret anyway: anyone holding it learns whether your server is up. Revoke it from the same panel.

Uptime Kuma: add a monitor of type HTTP(s), paste the URL, interval 60 s. BetterStack: create a monitor that alerts when the URL doesn't return 2XX, check frequency 1 minute.

A metrics export (OpenMetrics / Prometheus) is not available: series are read through the API (GET /api/v1/servers/{id}/metrics) with your own token.